Chapter 6

The Anatomy, Paleobiology, and Phylogenetic Relationships of the Hippopotamidae (Mammalia, Artiodactyla) from the Manonga Valley, Tanzania

TERRY HARRISON

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Materials .............................. . 3. Temporal and Geographic Distribution of Hexaprotodon harvardi 4. Craniodental Material . . . .

4.1. Cranium and Mandible 4.2. Upper Dentition 4.3. Lower Dentition

5. Postcranial Material . 5.1. Vertebrae .... 5.2. Pectoral Girdle and Forelimb 5.3. Hindlimb .......... . 5.4. Manus and Pes ....... . 5.5. Functional and Behavioral Implications of the Postcranium .

6. Taxonomy and Phylogenetic Relationships .......... . 6.1. Generic Affinities of the Manonga Valley Hippopotamid . 6.2. East Africa .. . 6.3. North Africa .... . 6.4. Europe and Asia . . .

7 Summary and Conclusions References . . . . . . . . .

.138

.138

.140

.141

.141

.148

.154

.158

.158

.160

.161

.163

.175

.175

.175

.179

.182

.183

.185

.186

TERRY HARRISON • Department of Anthropology, Paleoanthropology Laboratory, New York University, New York, New York 10003.

Neogene Paleontology o/the Manonga Valley, Tanzania, Volume 14 of Topics in Geobiology, edited by T. Harrison, Plenum Press, New York, 1997.

137

T. Harrison (ed.), Neogene Paleontology of the Manonga Valley, Tanzania© Springer Science+Business Media New York 1997

138 Chapter 6

1. Introduction

Hippopotamids are well represented in the Manonga Valley fauna, composing 23.4% of all large mammals, and they are second in importance only to bovids. Several species of hippopotamids are represented. The material from the Thole and Tinde Members, with the exception of a single postcranial specimen, can be assigned to Hexaprotodon harvardi, a large hexaprotodont hippopotamid, which is relatively common at late Miocene and early Pliocene sites in East Africa (Harrison, 1993). The remaining specimen from the nnde Member, an isolated phalanx, apparently belongs to a smaller species of Hexaprotodon. The taxo­nomic affinities of the hippopotamids from the overlying Kiloleli Member are more difficult to ascertain because so few specimens have been recovered. Even though they are most reasonably referred to Hex. harvardi, they differ in some minor details from the type material, and it is conceivable that they may represent a somewhat more progressive form. In addition to Neogene hippopo­tamids, several isolated teeth and some postcranial remains of Hippopotamus have been recovered from late Quaternary horizons. Apart from a brief discussion of their biochronological implications, these latter specimens are not included in the following analysis.

The aim of this chapter is to present a brief descriptive and comparative account of the hippopotamid material from Neogene sediments in the Manonga Valley. As the specimens are rather fragmentary, consisting primarily of isolated teeth and unassociated postcranials, they do not provide much new information about the anatomy of Hex. harvardi, which is otherwise well known from Lothagam and from sites in the Baringo basin. However, the Manonga Valley material is significant for two main reasons: (1) It serves to extend the geographic range of Hex. harvardi from southern Ethiopia, and northern and central Kenya, southward as far as northern Tanzania (the Baringo basin in Kenya, previously the southernmost locality with Hex. harvardi, is located almost 600 km northeast of the Manonga Valley), and (2) it provides a new focus for reassessing the paleobiology, taxonomy, and phylogenetic relationships of this species, which has received relatively little attention since it was initially described by Coryn­don (1977).

2. Materials

Almost three hundred hippopotamid specimens are known from the Ma­nonga Valley (Table I). There are no complete crania, and in fact, cranial and mandibular specimens are poorly represented. The collection comprises mostly isolated teeth (28%) and postcranials (67%). Hippopotamids have been recov­ered from all of the major stratigraphic units of the Wembere-Manonga Forma­tion (see Harrison and Verniers, 1993; Verniers, this volume, Chapter 2, for a detailed discussion of the stratigraphy), although they occur much less fre­quently in the Thole and Kiloleli Members than they do in the nnde Member, which has yielded almost 70% of the material described here (Table I). Material

Hippopotamidae 139

Table I. Number and Distribution of Hippopotamid Remains from the Manonga Valley

Cranial Mandibular Unit Localities fragments fragments Teeth Post-cranials Total

Kiloleli Ngofila4 0 0 0 3 3 Member Beredi South 1 & 3 0 0 2 1 3

Kiloleli 2-4 1 0 16 29 46

TInde TInde (1929) 4 0 19 22 45 Member

TindeEast 0 1 6 33 40 TIndeWest 3 1 24 96 124

Ibole Shoshamagai 2 2 0 9 1 12 Member Inolelo 1-3 2 0 8 16 26

Total 12 2 84 201 299

from the Thole and Tinde Members is clearly identifiable as Hex. harvardi, while the few specimens from the Kiloleli Member are provisionally assigned to the same species. An isolated phalanx from the Tinde Member apparently belongs to a smaller species of Hexaprotodon.

In addition, a small collection of hippo pot am ids from Kiloleli 2 was obtained from a yellow clay horizon at the top of the sequence. The bones are pale gray, with a yellowish tinge, and they are generally chalky and quite friable in nature. A similar assemblage has also been recorded from a horizon above the Kiloleli Member at Ngofila 2, where several ofthe bones exhibit signs of human activity in the form of cut marks. The material from Kiloleli 2 consists of some associated postcranial remains and several isolated teeth of a large hippopotamid. The stout metapodials, the broad unciform with a reduced styloid process, and the narrow, hypsodont lower molars are typical of Hippopotamus, and comparisons show that they are morphologically indistinguishable from modern Hip. amphibius. Although few other fossil mammals were recovered from this particular horizon at Kiloleli 2, several associated teeth have been identified as Phacochoerus sp. It is reasonable to assume that these bones are late Pleistocene or Holocene in age. The only other hippopotamid specimen recorded from the Manonga Valley is a fragment of a molar, probably of Hip. amphibius, from the mbuga clays at Kininginila. These clays are widely distributed throughout the Manonga Valley as superficial sediments, and they are estimated to be late Quaternary in age (see Harrison and Baker, this volume, Chapter 13).

The major portion of the hippopotamid material included in this study was collected between 1990 and 1994 by the Wembere-Manonga Paleontological Expedition (WMPE). This material is housed in the National Museums of Tanzania in Dar es Salaam. The remaining specimens were recovered in 1929 by Grace and Stockley at the site of Tinde, and this latter collection is housed in the Natural History Museum, London. Specimens from the National Museums of Tanzania are identified by their field numbers, which are prefixed by the letters WM (for Wembere-Manonga), while the London specimens are prefixed by the letter M. Comparisons with fossil hippos from other sites in Africa and Eurasia,

140 Chapter 6

as well as extant hippos, were made at the Natural History Museum, London, the Rijksmuseum van Natuurlijke Historie, Leiden, the Institut Paleontologic, Sabadell, the American Museum of Natural History, New York, and the National Museums of Kenya, Nairobi.

3. Temporal and Geographic Distribution of Hexaprotodon harvardi

Preliminary reports on the geology and paleontology of the Thrkana and Baringo basins in Kenya made reference to the occurrence of a previously undescribed primitive hexaprotodont hippopotamid from late Miocene and early Pliocene sediments (Patterson, 1966; Patterson et a1., 1970; Coryndon, 1970; Bishop et a1., 1971; Coryndon and Coppens, 1973; Maglio, 1974). Patterson et a1. (1970) identified this material as Hippopotamus (Hexaprotodonj sp. nov. A. A brief review of the fossil hippopotamids from the Turkana basin led Coryndon (1976) to identify a new species from Lothagam 1 and Kanapoi, which she provisionally referred to as sp. "D" nov. Later, Coryndon (1977) formally proposed the name Hex. haTVardi for this species. Additional specimens were subsequently described from sites of similar age in the Baringo basin (Coryndon, 1978a,b). To date. only preliminary accounts of the morphology of Hex. haTVardi have been published (Coryndon. 1977. 1978a.b). so the recovery of new material from the Manonga Valley provides an ideal opportunity to present a more comprehensive overview of the anatomy and taxonomic status of Hex. haTVardi.

Hexaprotodon haTVardi represents the earliest known member of the Hip­popotaminae (see section 6 for a fuller discussion of its phylogenetic status). It is restricted to the late Miocene and early Pliocene of East Africa. known primarily from sites in the Thrkana and Baringo basins of Kenya, dated at 7-4 Ma. The species is best known from the type site of Lothagam in northern Kenya. where an extensive series of well-preserved cranial and postcranial specimens has been recovered (Coryndon, 1977. 1978a). Based on faunal comparisons. the age of Lothagam is generally regarded as 5-7 Ma (Patterson et a1 .• 1970; Maglio. 1974; Smart. 1976; Behrensmeyer, 1976; Hill and Ward. 1988; Hill et al.. 1992. Hill. 1994). This has been confirmed by recent radiometric dates (Leakey et a1., 1996) which give an age of 4.72-6.24 Ma for the main fossil beds. Coryndon (1977). in her initial description of the species. also included material from Kanapoi. a site about 50 km south of Lothagam on the western side of Lake Turkana. The sediments are younger than those from Lothagam. and are esti­mated to be 3.9-4.2 Ma (Patterson. 1966. Maglio. 1974; Hill. 1994; Brown.1994; Leakey et al.. 1995). As noted by Coryndon (1977, 1978a). the sample from Kanapoi is somewhat more derived than the type material in several key craniodental characteristics (see section 6.2 for details). Leakey et a1. (1995) recently identified the hippopotamid from Kanapoi as Hex. cf. protamphibius. but detailed comparisons of the cranial and dental morphology confirms its closer affinities with Hex. haTVardi.

Hippopotamidae 141

The species has also been recorded from Baringo (Coryndon, 1978a,b). The best sample comes from the Lukeino Formation, dated at 5.6-6.2 Ma (Hill et aI., 1985,1986; Hill and Ward, 1988; Hill, 1994). These specimens are indistinguish­able from those from Lothagam. A small, but important, collection is also known from the earlier Mpesida Beds, dated at 6.4-7.0 Ma (Hill et aI., 1985, 1992; Hill and Ward, 1988; Hill, 1994). These specimens are morphologically very similar to Hex. harvardi from Lothagam and Lukeino, but they are slightly larger in size (the cheek teeth have an occlusal area that is 16.6% larger, and several specimens exceed the 95 % confidence limits for the combined sample from Lothagam and Lukeino). Coryndon (1978b) also noted a general similarity between the speci­mens from Lukeino and those from the Toluk and Aterir Beds, which date from about 4.5-5.0 Ma (Coryndon, 1978a; Hill, 1994), although the collections are too fragmentary to be able to attribute them with certainty to Hex. harvardi. It is also possible that Hex. harvardi may be represented in the lower Chemeron Forma­tion, which is estimated to be 3.7-5.6 Ma (Coryndon, 1978b; Hill et aI., 1985; Hill, 1994). A few isolated teeth of a medium-size hippopotamid, almost cer­tainly attributable to Hex. harvardi, have been reported from the Ngorora Formation, which is estimated to be 9.0-12.3 Ma (Bishop and Chapman, 1970; Maglio, 1974; Bishop and Pickford, 1975; Pickford, 1978a; Coryndon, 1978a,b; Hill et aI., 1985; Geze, 1985; Hill and Ward, 1988; Hill, 1994). However, Pickford (1983) has argued that the provenience of these specimens is doubtful, and that they may be derived from the Mpesida Beds or the Lukeino Formation.

Further north, in Ethiopia, Hex. harvardi has been provisionally identified from the Adu Asa and Sagantole Formations of the Middle Awash Valley (Kalb et aI., 1982a,b,c). These sediments are estimated to be 6-4 Ma (Kalb et aI., 1982a,b,c; Kalb and Jolly, 1982; Kalb & Mebrate, 1993; Kalb, 1993), consistent in age with Hex. harvardi sites in northern Kenya.

4. Craniodental Material

Unfortunately, cranial and mandibular specimens are not well represented in the collections from the Manonga Valley, and most of the craniodental specimens consist of isolated teeth (see Table I). All of the permanent dentition, with the exception of P l , is represented, however, and the collections also include examples of upper and lower deciduous premolars. A catalog ofthe craniodental specimens from the Manonga Valley is presented in Table II, and a list of standard measurements is given in Table III. A description of the new material from the Manonga Valley is presented in the following sections, along with a discussion of the general morphology, comparative anatomy, and taxonomy of Hex. har­vardi.

4.1. Cranium and Mandible

Few cranial specimens have been recovered from the Manonga Valley. WM 056/90 comprises a left premaxilla of a juvenile individual in which the three

142 Chapter 6

Table II. List of Hippopotamid Craniodental Specimens from the Manonga Valleyu Tindeb

M44683

M44684 M44685 M44686 M44687 M44688 M44689 M44690 M44691 M44692 M44693 M44709 M44710 M44711 M44712 M44713 M44714 M44715 M44716 M44717 M44718a-b M44719 M44712

TindeWest WM036/90 WM056/90 WM245/90 WM 268/90 WM 283/90 WM469/90 WM470/90

WM481/90 WM 557/90 WM 558/90 WM814/90 WM087/92 WM 113/92 WM 222/92 WM421/92 WM 521/92 WM 547/92 WM572/92 WM 577/92 WM621/92 WM 763/92 WM 1899/92 WM 1905/92 WM 1906/92 WM105/94 WM910/94 WM918/94 WM924/94

Right maxilla of juvenile individual with canine erupting and roots of pI and dP2-4, and the anterior alveolus for Ml

Right Ml or M2 in maxilla fragment, heavily worn Right M2, moderately worn Left M3, lightly worn LeftM1

Left dP2' apex of main cusp lacking Right p2, worn. distal margin missing Right maxillary fragment with distobuccal portion of Ml and alveolus of M2 Left M2• mesiolingual portion of the crown lacking Left maxillary fragment of juvenile individual with M2 preserved in crypt Lower canine. fragment Upper canine. fragment Lower canine. tip of crown only Lower canine, fragment Lower canine. fragment Lower canine. fragment Lower canine. fragment Upper canine. fragment Lower incisor. fragment Lower incisor. fragment Lower canine. two conjoining fragments Lower incisor. fragment Lower incisor. fragment

Right M2. fragment Right premaxilla of sub adult individual with the 11- 3 just erupting Right mandibular fragment. edentulous with roots of M2-3 Right p3. distolingual portion of crown only Lower canine. fragment Right M3. distal portion of crown only. heavily worn Right MI , missing distal one-third and mesiolingual corner of crown. moderately

worn Right Ml in maxillary fragment. moderately worn Lower molar fragment Right P2 or P3 Lower incisor. worn Left p3. lingual accessory cuspule and portion of main cusp only Left p3. lingual heel of crown only, unworn Molar fragment Left dP2• mesial portion of crown only. slightly worn Upper molar fragment. probably Ml Right p4. unworn Right dP 4' distobuccal portion of crown only Upper canine. fragment Left M3• distal portion of crown only. heavily worn Lower canine. fragment Left P2 or P3, slightly damaged Right dP2. distal portion of crown only Right lower incisor, possibly 13• moderately worn Left p3, slightly damaged mesially Left dP2. distal portion of crown only. moderately worn Lower canine fragmen Left premaxilla of immature individual with 11- 2 erupting. root of dp3

(continued)

Hippopotamidae 143

Tinde East WM 319/90 WM 326/90 WM442/90 WM445/90 WM 695/92 WM 1854/92 WM 070/94

Kiloleli 2 WM 594/92 WM 852/92 WM 853/92 WM 855/92a WM 855/92b WM 855/92c WM 855/92d WM 1281/92 WM 1283/92 WM 394/94 WM 399/94

Kiloleli 3 WM 753/90 WM 755/90 WM 761/90 WM 794/90

Kiloleli 4 WM656/90 WM 808/90

Shoshamagai 2 WM 1169/92 WM 1780/92

WM 1810/92 WM 1811/92 WM 1813/92 WM 1815/92 WM 1818/92 WM 1932/92 WM 198/94 WM 286/94

WM 299/94

Inolelo 1 WM 1000/92 WM 1001/92 WM 1064/92 WM 1119/92 WM 1121/92 WM 1150/92 WM 1151/92 WM 140/94

Table II. (Continued)

Left M3, mesiobuccal corner of crown missing, moderately worn Lower canine, fragment Lower canine, 3 fragments Right upper canine, medial portion of apex only Right mandibular fragment, edentulous with the roots ofMl _2 Right M2, missing mesiolingual corner of crown, heavily worn Lower canine, fragment

Left p3, lacking mesial portion of crown, moderately worn Right M3, mesiobuccal corner missing, unworn Right M1 or M2, heavily worn Right p4, distobuccal fragment of crown only, unworn Left P2 or P3' distal fragment of crown only Right p 4' distobuccal portion of crown only Right lower incisor, probably I3, moderately worn Upper molar fragment Upper incisor, probably right I2, tip of crown only Lower canine, fragment Tip of upper incisor and base of crown of lower incisor

Left lower canine, fragment Right upper incisor, probably I1 Right M2, distal two-thirds of crown only, unworn germ Right maxillary fragment, edentulous with the roots of M l - 3

Left dP3' distal portion of crown only, worn Left M1 or M2, moderately worn

Upper canine, fragment Left maxillary fragment with dp3, roots of dP4 , alveoli of dp2 and M1, dentition

moderately worn Left M2, moderately worn Left M2, distal moiety of crown only, moderately worn I3, heavily worn Right dP 4' distal portion of crown only, moderately worn Right dp2, distal portion of crown only, tip of main cusp incomplete, unworn Upper molar fragment Left M3, moderately worn Left maxillary fragment with distal portion of dp4, heavily worn, M1, moderately worn, and the partially preserved crowns ofP4 and M2 preserved in crypts Tip of upper incisor, probably left I2

Left M2, moderately worn Left upper canine, fragment Left I3, very worn Left p3, distal portion of crown only, slightly worn Right p1 in maxillary fragment, worn Left M2, mesial moiety of crown only, heavily worn Molar fragment Lower canine, fragment

(continued)

144

Inolelo 3 WM664/94

WM669/94

Beredi South 1

Chapter 6

Table II. (Continued)

Left maxillary fragment with base of erupting canine. roots ofPl, base of dPz• and pZ exposed in crypt

Left Pz or P3

WM 1514/92 Left lower incisor. possibly 11. tip of crown missing. slightly worn

Beredi Soutlt 3 WM 1670/92 Right upper canine. fragment

a Specimens from the Natural History Museum in London have accession numbers prefixed by M. Specimens from the National Museums of Tanzania in Dar es Salaam are unaccessioned. and are referred to here by their field numbers only (preflxed by WM for Wembere-Manonga).

bAll of the specimens in the Natural History Museum in London were collected in 1929 by Grace and Stockley at the site of Tinde. The field notes of Grace and Stockley establish that they collected material from both Tinde West and Tinde East. but the two samples cannot be differentiated.

TableID. List of Dental Measurements of Hippopotamids from the Manonga Valleyu

Upper dentition

Max

WM056/90 11 13.0(-) IZ 14.2

WM 755/90 I 18.6 WM 1064/92 15.4 WM 1283/92 22.2

AP TR

WM 577/92 C1 46.1(-) WM 1001/92 C1 48.3(-) 54.8(-) WM 1670/92 C1 36.4(-) 44.3(-) M44709 C1 33.8(-) 51.0(-)

MD BL BLMes BL Dist HT

WM421/92 dPz 24.4 WM 1818/92 dPz 16.4 16.4 WM 1905/92 dPz 15.0 15.0 WM 1780/92 dP3 35.2 21.3 21.3 WM 1121/92 pI 22.6 15.3 M44689 pZ 38.0 26.6 WM594/90 p3 30.5(-) WM 1119/92 p3 30.5(-) WM 105/94 p3 26.9 WM 547/92 p4 29.3 36.7 WM481/90 Ml 46.2 41.6 41.6 40.1 M44687 Ml 44.0 38.1 37.6 38.1 M44684 M1IZ 42.3 WM 761/90 MZ 47.2 46.0 WM 1810/92 MZ 48.2 42.0 37.5 WM 1811/92 MZ 37.7 WM 1854/92 MZ 53.0 47.3 M44686 M3 48.2 49.9 49.9 43.6

(continued)

Hippopotamidae

Max

WM814/90 19.1 WM 855/92 18.0 WM 1514/92 18.0 WM 1906/92 16.9 M44716 34.5 M44717 38.4 M44719 22.4 M44720 23.7

AP

WM 326/90 C1 66.0(-) WM 753/90 C1 56.8(-) WM 763/92 C1 51.5(-) WM070/94 C1 60.6(-) M44693 C1 61.3(-) M44710 C1 50.0(-) M 44711 C1

M44718 C1 57.2(-)

MD

M44688 dPz 29.1 WM656/90 dP3 WM 558/90 P Z/3 37.2 WM 1899/92 PZ/3 37.0(-) WM470/92 Ml WM808/90 M1/Z 51.5 WM 853/92 M1/ Z 47.3 WM 1000/92 Mz 56.0 WM 1150/92 Mz M44685 Mz 50.4 M44691 Mz 45.5 WM319/90 M3 66.6 WM852f92 M3 72.7 WM198/94 M3 69.2

Table m. (Continued) Lower dentition

TR

35.2(-) 30.1(-)

34.5(-) 37.4(-) 32.5(-) 34.6(-) 34.9(-)

BL BLMes

14.7 14.7

23.3 21.2 24.2 19.4

30.0(-) 34.1 34.1 32.2 31.0 40.7 40.5

37.4 34.9 34.9

40.0

BLDist

13.0(-) 17.7 23.3 24.2

33.0 32.2 40.7

33.2 38.5 34.1 38.9

HT

37.8

31.3

41.2 47.6 41.5

145

Abbreviations: AP. maximum anteroposterior length; BL. maximum buccolingual breadth; BL Dist. buccolingual breadth of distal moiety of crown; BL Mes. buccolingual breadth of mesial moiety of crown; HT. maximum height of crown; M. accession number prefIx for specimens housed in the Natural History Museum. London; Max. Maximum diameter; MD. maximum mesiodistallengtb; TR. maximum transverse width; WM. fIeld number prefIx for specimens housed in the National Museums of Tanzania; (-) minimum value.

permanent incisors are just beginning to erupt (Fig. 1). The premaxilla is weathered and slightly abraded, having been exposed on the surface for some time prior to collection. The specimen is morphologically very similar to the premaxilla in the type specimen of Hex. harvardi from Lothagam (KNM-LT 4). The Manonga Valley specimen is somewhat smaller and less robust than the latter, but this may be accounted for by their difference in age; the Lothagam specimen represents the skull of an adult with advanced dental wear.

The palatal portion of the premaxilla is relatively flat, and the two halves of the bone evidently met coronally in the midsagittal line at an obtuse angle. As

146 Chapter 6

c d

f

n

-FIGURE 1 . Craniodental specimens of Hex. harvardi from the Manonga Valley. (a) WM 056/90. left premaxilla with 11- 3, palatal view; (b) WM 558/90, right P2 or P3, ligual view; (c) WM 1121/92, right PI. lingual view; (d) 547/92 , right p4, occlusal view; (e) WM 481/90, right M\ occlusal view; (f) 1854/92, right MZ, occlusal view; (g) WM 481/90, right M1, lateral view; (h) WM 481/90, right M\ medial view; (i) WM 853/92, right M1/Z, occlusal view; 0) WM 808/90, left M1I2, occlusal view; (k) WM 1000/92, left Mz, occlusal view; 0) WM 319/90, left M3, occlusal view; (m) WM 853/92, right M1!Z, buccal view; (n) WM 808/90, left M1/Z, buccal view; (0) WM 1000/92, left Mz, buccal view; (p) WM 319/90, left M3, buccal view. Scale bars = 10 mm. Top scale bar for a-h, bottom scale bar for i-po

is typical of Hexaprotodon (and also Choeropsis), the premaxillae are fused along their entire length in the midsagittal line, and there is no separation anteriorly, as is commonly seen in Hippopotamus (Colbert, 1935; Coryndon, 1977; Steunes, 1989) . Posteriorly, a remnant ofthe anterolateral margin of the incisive foramen is preserved. It is located close to the midline of the palate, just posterior to the central incisors. Lateral to the large opening for the incisive foramen is a distinct groove, presumably the anterior continuation of a smaller subsidiary incisive canal. The anterior margin of the premaxilla is broad and straight. It curves laterally around the root of 13 , and then is separated by a shallow concavity from the widely flaring maxilla, which is greatly expanded to accommodate a massive upper canine. The maximum breadth of the palate anteriorly can be estimated to have been 125 mm. The facial portion of the premaxilla is quite steep anteriorly and inferiorly, but it flattens out superiorly to produce a more gently sloping lateral margin to the nasal aperture. The nasal aperture has a broad

Hippopotamidae 147

U-shaped margin inferiorly, and it approaches very close to the alveolar margin of the incisors (the height of the nasoalveolar clivus in the midline is only 15.5 mm).

A second premaxillary fragment (WM 924/94) from Tinde West confirms the hexaprotbdont nature of the dentition. It is a left premaxilla of an immature individual preserving the roots of 11 and d13, and the broken crown of 12, whiCh was just beginning to erupt. The two permanent incisors are situated side by side in the front of the jaw, while the dI3 is positioned somewhat more posteriorly.

Five additional cranial specimens preserve portions of the maxilla. M 44683 consists of a right maxillary fragment of a juvenile preserving the tip of the canine, just beginning to erupt, and the roots of pi and dP2--4; WM 794/90 is an edentulous right maxillary fragment with the roots of Ml - 3; WM 1780/92 is a left maxilla of a juvenile with dp3; WM 286/94 is a left maxilla with dp4-M\ and portions of the crowns of p4 and M2 retained in their crypts; and WM 664/94 is a left maxillary fragment of a juvenile individual preserving the base of the canine, the roots ofP\ the base of the crown of dp2, and the unerupted crown of p2 exposed in its crypt.

The palate appears to have been relatively deep and concave to V-shaped in section. The estimated breadth of the palate at the level of dp3 in WM 1780/92 is 73 mm. In the juvenile specimens the facial aspect of the maxilla is steep and slightly concave above dp2-3. The infraorbital foramen is located 45 mm above the alveolar margin of the maxilla, which indicates a relatively deep face when compared with Hip. amphibius of similar dental age, but similar to G.liberiensis. The maxilla is stoutly constructed, with a solid corpus of bone accommodating the roots of the cheek teeth. The anterior root of the zygomatic arch originates very low down on the face, close to the alveolar margin of the maxilla, as in Choeropsis. In Hip. amphibius the zygomatic arch originates higher on the face, in association with a deeper lower face and more elevated orbits.

Cranially, Hex. haTVardi shares with Choeropsis a number of features that can reasonably be inferred to be plesiomorphic for the Hippopotaminae: (1) There is a tendency for the median suture of the premaxillae to be fused anteriorly; (2) the muzzle is relatively short in relation to the length of the neurocranium; (3) the orbits are placed laterally (not elevated superiorly as in Hip. amphibius) and located about midway along the length of the cranium; (4) the anterior root of the zygomatic arch is situated low on the face, at or below the level of the infraorbital foramen (in Hip. amphibius the infraorbital foramen is situated lower on the face relative to the root of the zygomatic arch); (4) the nasals are expanded posteriorly; (5) lacrimal and nasal bones are separated by a well-developed antorbital process of the frontal; and (6) the nuchal region does not rise superiorly much above the level of the frontal bone.

The mandible is known only from WM 695/92 and WM 245/90, but these are too incomplete to provide any detailed morphological information. The mandibular corpus appears to be relatively robust, and the anterior root of the ascending ramus originates opposite M3.

148 Chapter 6

4.2. Upper Dentition

4.2.1. Incisors

WM 056/90, a right premaxilla of a subadult individual, preserves the crowns of 11_13 , in an early stage of eruption (Fig. 1). The crowns are enamel-covered and unworn, but they are weathered and badly abraded, so details of their morphol­ogy cannot be determined. All three incisors appear to be sub equal in size, evenly spaced, and arranged in a gently curving arc in the premaxilla. The crowns are relatively small in relation to the size of the premaxilla, but the incisors would presumably have increased in diameter with continued growth. 11 is subcircular in cross section. 12 is more elliptical, with the mesiodistal diameter greater than the buccolingual diameter. 13 is still partially retained within the crypt, so its cross-sectional shape cannot be ascertained. A similar morphological pattern can be reconstructed from the anterior dentition in a juvenile premaxilla, WM 924/94.

WM 056/90 and WM 924/94 definitively establish that the Manonga Valley hippopotamid was hexaprotodont. In terms of their relative size, the upper incisors are similar to those of Hex. harvardi from Lothagam and Lukeino, in which all three incisors are sub equal (see Table IV for comparative data on upper incisors). An almost complete skull of Hex. harvardi from Kanapoi (KNM-KP 8529), however, is more derived in having an 11 that is distinctly larger than the other two incisors. Subequal upper incisors are also characteristic of Hex. sivalensis from the early Pliocene of the Siwalik Hills, but there is a tendency in this species for 13 to be somewhat reduced. The only other hexaprotodont

Table IV. Comparison of Uppper Incisors in Fossil and Extant HippopotBlllids

Species

Hexaprotodon harvardi Lothagam & Lukeino Manonga Valley Kanapoi

Hexaprotodon sivalensis sivalensis

Trilobophorus afarensis Hexaprotodon protamphibius

turkanensis Hexaprotodon aethiopicus Hexaprotodon karumensis

Koobi Fora (Upper Burgi) Koobi Fora (Upper KBS)

Hippopotamus gorgops Hippopotamus amphibius Choeropsis liberiensis

Number of incisors

Hexaprotodont Hexaprotodont Hexaprotodont Hexaprotodont

Hexaprotodont Hexaprotodont

Tetraprotodont

Tetraprotodont Diprotodont Tetraprotodont Tetraprotodont Tetraprotodont

Relative proportions of incisors

11-3 subequal (n = 1. 80:100:84)a 11-3 subequal 11 larger.12-3 subequal 11-2 subequal. 13 slightly smaller (n = 1.

89:100:77) 11-3 subequal 11- 3 sub equal

11-2 subequal (n = 2. 100:93)

11 much larger than 12 (n = 1. 100:72)

11-2 sub equal (n = 1. 88:100)

11-2 subequal (n = 29. 100:80)b

11- 2 subequal (n = 9. 98:100)b

a Numbers in parentheses represent the relative size (maximum mesiodistal diameters) of the incisors. The largest incisor is equal to 100. They are average values arranged in sequence starting with 11. n is the sample size.

b Data from Pavlakis (1987).

Hi ppopotamidae 149

hippopotamids from East Africa are Hex. protamphibius turkanensis from the lower part of the sequence in the Turkana basin, Hex. coryndonae and Trilobo­phorus ajarensis from the Afar region of Ethiopia, and possibly also Hex. imagunculus from the Western Rift. I have no detailed information on the relative proportions of the upper incisors in these species, but they were apparently subequal, at least in Hex. protamphibius turkanensis and Trilobophorus ajaren­sis (Geze, 1985). Most other Plio-Pleistocene hippopotamids from East Africa are distinguished from Hex. haTVardi in having a tetraprotrodont pattern with two subequal incisors (Geze, 1985; Harris, 1991, Table IV). However, Hex. karumensis from the upper Burgi and lower KBS Members at Koobi Fora is further derived in having a tetraprotodont arrangement in which the lateral incisor is consider­ably smaller than the central incisor, and this subsequently gave rise to a diprotodont form in the upper KBS Member (Harris, 1991).

The Manonga Valley collections include several isolated upper incisors, but identification of their serial position in the jaw has proved problematic. WM 755/90 and WM 1283/92 are moderately worn, and probably represent 11 and 12 ,

respectively. WM 1064/92 and WM 1813/92, both possibly attributable to 13, are at a more advanced stage of wear. Only the apical portions of the incisors are preserved. The enamel, which is restricted to a continuous strip along the buccal face of the crown, is quite thick, and is marked by a series of faint longitudinal striae. The incisors exhibit a marked curvature toward the mesial and lingual aspects. In cross section, the crowns are rectangular to oval in shape, with the buccolingual diameter being slightly greater than the mesiodistal diameter. In intermediate stages, wear is restricted to the lingual or distolingual face of the crown. This produces a relatively flat or mesiodistally convex occlusal plane, bordered buccally by a raised enamel margin. This ensures that, with continued wear, a sharp incisive edge is maintained. As noted by Coryndon (1978a), this type of wear, produced by tip-to-tip occlusion with the lower incisors, is typical of Hexaprotodon. However, in more worn specimens, such as WM 1064/92 and WM 1813/92, in which the enamel surface has been entirely lost through attrition, the occlusal surface is biconvex or conical in shape.

Morphologically and metrically the upper incisors from the Manonga Valley are consistent with those of Hex. haTVardi from Lothagam. They share the following characteristics: (1) hexaprotodonty, (2) subequality in size, and (3) tip-to-tip occlusion with the lower incisors.

4.2.2. Canine Partially preserved upper canines have been recovered from Tinde West,

Tinde East, Shoshamagai 2, Inolelo 1, and Beredi South 3 (see Table III). The morphology is typical of upper canines of Hexaprotodon. They bear a deep V -shaped groove posteriorly, bordered medially and laterally by angular margins. In cross section the upper canine is mediolaterally wider than long, with a flattened anterior face that is narrower than the posterior face. The canine has a convex medial margin, so that the crown curves outwards laterally when seen in ventral view. The enamel is distributed only on the posterior and lateral aspects of the tooth, and its surface is marked by fine striae, rather than the coarse

150 Chapter 6

ribs characteristic of Hippopotamus. The upper canines wear obliquely antero­posteriorly, at an angle of approximately 1350 to the anterior plane, so that the anterior margin of the crown is less elevated above the alveolar plane than the posterior margin.

In comparison with Hip. amphibius the canines appear to be relatively large. However, given the high degree of variability (especially with canines that continue to grow throughout life) and sexual dimorphism in canine size in hippopotamids, this is a difficult feature to compare, especially with the small sample sizes available for fossil hippopotamids. Some indication of relative canine size, however, can be gained by relating maximum diameter of the base of the upper canine to the mesiodistal length of M2. In a sample of 27 skulls of Hip. amphibius, the upper canine was found to be mesiodistally shorter than M2 (with an index less than 100) in all individuals, except for a single specimen with an index ofl07 (range = 55-107; mean = 76). By contrast, the modern pygmy hippo, Choeropsis liberiensis, has relatively larger upper canines, being usually somewhat mesiodistally longer than M2 (n = 11; mean index = 110; range = 87-133). Comparative data for Hex. harvardi (n = 5; mean index = 108; range = 84-128) confirm that this species has relatively large upper canines. Although there are no upper canines directly associated with molars from the Manonga Valley, comparisons of isolated teeth suggest that the canines were relatively large, similar in size to those from Lothagam and Mpesida.

In sum, the upper canines from the Manonga Valley are identical to those of Hex. harvardi from Lothagam and Lukeino. They share the following charac­teristics that are typical of Hexaprotodon (and Choeropsis): (1) They are rela­tively large in size; (2) there is a deep groove along the posterior aspect of the crown; and (3) the enamel surface is finely striated.

4.2.3. Premolars pl is represented by a single isolated specimen, WM 1121/92, from Inolelo 1

(Fig. 1). The crown is mesiodistally long and narrow, and it exhibits a slight degree ofbuccolingual waisting. It has a single main cusp, which, although worn, was evidently quite tall, with a slight lingual curvature toward its apex. The cingulum is narrow and ill-defined, but it surrounds most of the base of the crown.

This specimen is similar in morphology to those from Lothagam and Lukeino, although it is somewhat smaller. However, it is distinctly larger than the pl in later East African hippopotamids, which is very small, and tends to be shed with advancing age. Moreover, maxillary fragments from Tinde (M 44683) and Inolelo 3 (WM 664/94) preserve the roots of pl, and judging from their size they would have accommodated a sizable tooth (at least 29 mm long in WM 664/94).

Coryndon (1977, 1978a,b) has suggested that Hex. harvardi is characterized by having a double-rooted pl, a feature also seen in Hex. sivalensis, and in hippopotami from the Holocene of Madagascar (Steunes, 1981, 1989). Although the root ofthe tooth is still embedded in bone in WM 1121/92, from the contour of its cervix, as well as from its exposed tip, it appears that only a single root was present. M 44683 and WM 664/94, however, do preserve the roots of a double-

Hippopotamidae 151

rooted pl. Similar variability in this feature can also be shown to occur in samples of Hex harvardi from other sites, such as Lothagam and Kanapoi.

p2 is a long and narrow tooth, with a single main cusp. The cingulum is continuous around the base of the crown. Arising from the main cusp is a distal crest, which bears several prominent cuspules. p3 is represented by a number of fragmentary specimens, which together allow a composite description of the tooth. The crown is triangular in shape, being broadest in its distal moiety, owing to the development of a prominent distolingual heel. There is a single main cusp, which is tall, lingually recurved, and triangular in cross section. Three crests descend from the apex of the main cusp. The mesial crest arcs basally to terminate at a well-developed cingular shelf. The two distal crests and the distal basin of the crown bear numerous accessory cuslmles. A continuous cingulum passes around the base of the crown, but it is especially well developed along the lingual margin. p4 is a bicuspid tooth, with a well-developed accessory cusp that is only slightly less elevated than the main cusp (Fig. 1). The crown is broader than long, and elliptical to triangular in shape. A strong cingulum completely surrounds the base of the crown. Accessory cuspules are variably developed. The occlusal area ofP4 falls within the 95% confidence limits ofthe combined samples from Lothagam and Lukeino (Table V). Although the upper premolars are variable in size, shape, and structure in fossil hippopotamids, the specimens from the Manonga Valley are similar to those from Lothagam and Lukeino in being large, robust, polycuspidate teeth.

As noted by Coryndon (1978a), the length of the upper premolar series is subequal to or exceeds that of the molar row in Hex. harvardi. However, as pl is commonly lacking in hippopotamids and because fossil material is rarely com­plete enough to obtain data on the lengths of the premolar and molar series, a

Table V. Size of the Cheek Teeth of Hex. harvardi from Lothagam, Lukeino, and Manonga Valley

Occlusal area (mesiodistal length x buccolingual breadth)

Lothagam and Lukeinou

N Mean Range S.D. Manonga Valley b

p3 10 1280 994-1560 183 p4 22 1233 985-1726 176 1075 Mt 13 1630 1318-1969 206 1676,1921 M2 16 2314 1688-3037 400 2024,2507 M3 18 2221 1353-2586 297 2405

P3 4 1004 879-1151 104 867,895(-) P4 14 1176 874-1738 219 Mt 13 1341 1035-1673 179 M2 20 1828 1371-2286 189 1752,1759,2279 M3 17 2319 1789-2785 239 2271(-), 2828(-), 2768

Abbreviations: N, sample size; S.D., standard deviation; (-) minimum value. a Data from Lothagam and Lukeino is combined. b Data from Manonga Valley represents individual specimens.

152 Chapter 6

more practical measure of relative upper premolar size has been devised. The index is calculated as follows: The sum of the mean lengths of individual teeth for the p2-4 series is divided by the sum of the mean lengths of individual teeth in the Ml - 3 series, expressed as a percentage. In Hex. harvardi, this index is 81, which confirms that the premolar series is relatively long when compared with most other hippopotamids (Hex. sivalensis = 75; Hex. karumensis = 73; Hex. aethiopicus = 68; Hip. gorgops = 65; Hip. major = 69; Hip. amphibius = 68; C. liberiensis = 83). An alternative measure of the relative size of the premolars is given by the total occlusal area of the upper and lower P3 and P4 expressed as a percentage of the total occlusal area of the upper and lower molars. In Hex. harvardi the P3 and P4 have an occlusal area that is 40% of that of the molars, whereas in all other hippopotamids P3 and P4 are much smaller, with an area that is only 30 ± 5% of the molar occlusal area (Table VI).

In summary, the upper premolars from the Manonga Valley share the follow­ing characteristic features with Hex. harvardi from Lothagam: (1) They are relatively large in size and robust, with pustulate enamel; (2) p 1 is large, and commonly double-rooted; and (3) p4 is bicuspid, with a well-developed acces­sory cusp.

4.2.4. Molars

The upper molars are broad and low-crowned, with a prominent cingulum that entirely encircles the base of the crown (Fig. 1). The four main cusps are low and conical in shape. In lightly worn molars, each cusp bears a triangular-shaped exposure of dentine, but as wear advances this becomes an open trefoil shape typical of Hexaprotodon (Coryndon, 1978a). The hypsodonty index (height of the crown x 100/buccolingual breadth of the crown) cannot be calculated for any of the complete upper molars from the Manonga Valley because of their advanced stage of wear. However, the height can be measured in two partially preserved unworn upper molars (WM761/90 and WM 1810/92), and in both of these specimen the hypsodonty index can be estimated to have been less than 90. These values are comparable to those seen in Hex. harvardi, and in other species of Hexaprotodon from East Africa, which have a general range of 70-100. Choeropsis is similar to Hexaprotodon in having brachyodont upper molars; the hypsodonty index rarely exceeds 100. By contrast, Hip. amphibius has upper molars that are distinctly more hypsodont, with an index that may exceed 120.

The upper molars from the Manonga Valley are comparable in overall size to those from Lothagam and Lukeino, and in terms of their occlusal areas (mesio­distal length x buccolingual breadth) they all fall within the 95 % confidence limits for the combined sample from Lothagam and Lukeino (Table V). Compared with other hippopotamids, the upper molars are similar in size to Hex. sivalensis, slightly larger than Hex. protamphibius, and somewhat smaller than those of Hex. karumensis, Hip. amphibius, Hip. major, and Hip. gorgops. They are considerably larger than the dwarf forms Hex. imagunculus and Hex. aethiopicus, as well as the modern pygmy hippopotamus, C. liberiensis (Ta­ble VI).

Hippopotamidae

Table VI. Size and Proportions of the Cheek Teeth in Fossil and Extant Hippopotamidsa

ABC D

Hex. harvardi 4,693 11,653 100 100 Hex. sivalensis 3,398 12,062 72 104 Hex. protamphibius 3,029 9,620 65 83 Hex. karumensis 3,922 13,809 84 119 Hex. aethiopicus 1,916 7,112 41 61 Hex. imagunculus 2,235 6,860 48 59 Hip. gorgops 4,511 17,151 96 147 Hip. major 4,512 16,161 96 139 Hip. amphibius 4,209 13,444 90 115 C. liberiensis 1,101 3,510 23 30

153

E

40:100 28:100 31:100 28:100 27:100 33:100 26:100 28:100 31:100 31:100

a A, mean combined occlusal area (mesiodistal length x buccolingual breadth) of upper and lower P3 and P4 (mm2).

B, mean combined occlusal area of upper and lower molars (mm2). C, mean combined occlusal area of upper and lower P3 and P4 expressed as a percentage of that in Hex. harvardi. D, mean combined occlusal area of upper and lower molars expressed as a percentage of that in Hex. harvardi. E, mean combined occlusal area of upper and lower P3 and P4 expressed as a ratio of combined occlusal area of upper and lower molars. Source of data: Harrison (unpublished), Pavlakis (1987), Harris (1991), Faure (1985).

Another interesting characteristic of Hex. harvardi is the moderate size differential between the upper molars. Their relative size (occlusal area), ex­pressed as a ratio of the largest of the teeth, and arranged in sequence from M1 to M3, is 70:100:96. Comparative data for other hippopotamids are as follows: Hex. sivalensis, 66:96:100; Hex. protamphibius, 68:100:98; Hex. karumensis, 65:95:100; Hex. aethiopicus, 77:100:93; Hex. imagunculus, 50:91:100; Hip. gor­gops, 50:83:100; Hip. amphibius, 73:100:97; Choeropsis liberiensis, 61:99:100. In this respect, Hex. harvardi is similar to Hex. protamphibius and Hip. am­phibius, with most other hippopotamids having a somewhat more pronounced size differential between the upper molars.

In summary, the upper molars from the Manonga Valley are consistent in size and morphology with those of Hex. harvardi from Lothagam and Lukeino. They are distinguished from those of other species of Hexaprotodon from East Africa (but generally similar to Hex. sivalensis from Asia) by the following combination of features: (1) the upper molars are larger in size (although they tend to be slightly smaller than those of Hex. karumensis); (2) the crowns are more brachyo­dont; and (3) the size differential between the molars is less pronounced (except for Hex. protamphibius).

4.2.5. Deciduous Dentition

The dP2 is represented by four fragmentary specimens (WM 421/92, WM 1818/92, WM 1905/92, and WM 910194). The narrow, elongated crown has a distinct buccolingual waisting midway along its length. There is a single main cusp located in the midline of the crown, somewhat closer to the mesial than to the distal end of the tooth. Originating from the apex of the main cusp are sharp mesial and distal crests. The latter gives rise to a small, but prominent, accessory cusp, situated just to the buccal side of the midline. The basal cingulum is

154 Chapter 6

continuous around the distal, distobuccal, and lingual margins of the crown, but it is poorly developed on the mesiobuccal face. A well-preserved dP3 is associ­ated with the partial maxilla of a juvenile individual from Shoshamagai 2 (WM 1780/92). The crown is elongated and triangular in shape, being broadest distally and narrowing mesially. It is a molariform tooth, with three main cusps and a prominent mesial accessory cuspule. The protocone is a large, conical cusp situated in the midline of the crown, slightly toward the mesial end of the tooth. The two distal cusps are sub equal and transversely aligned. They are lower and less voluminous than the protocone. The cingulum is well developed, forming a narrow ledge that almost entirely surrounds the base of the crown. A small tubercle is located on the cingular shelf between the distolingual cusp and the protocone. The dp3 is very similar to those from Lothagam, although it is slightly smaller and relatively narrower.

4.3. Lower Dentition

4.3.1. Incisors

The lower incisors have low, stout, and conical crowns. The enamel covering is fairly thin, and is either smooth or finely crenulated. The base of the crown is bordered by an irregular but well-defined cingulum on its mesial, lingual, and distal aspects. The root is extremely long, and circular to elliptical in cross section. In early stages of wear a flat facet is cut obliquely down onto the lingual face of the crown, to produce a broad, chisel-shaped apical cutting edge. As wear progresses, and the incisor continues to grow, this facet gradually extends onto the root. In more aged individuals the enamel-covered portion of the crown is completely obliterated by wear, and all that remains is a stout cylinder of dentine with a smoothly polished, wedge-shaped tip. This continued growth also ex­plains why the maximum diameter of the largest incisor from the Manonga Valley is over twice that of the smallest incisor. In Hip. amphibius, aged adults commonly have lower central incisors that are more than three times the diameter of those of subadult individuals.

The similarity in size of the individual lower incisors from the Manonga Valley in which some enamel is still retained suggests that they may have been of relatively uniform size in the mandible. Subequal incisors are also found in mandibular specimens from Lothagam and Lukeino. Among other hexaproto­dont hippopotamids from East Africa, only Trilobophorus afarensis has sub equal lower incisors (Gaze, 1985). Interestingly, a lower jaw of Hex. harvardi from Kanapoi (KNM-KP 1) is more specialized in this regard in having a central incisor that is quite a bit larger than the two lateral incisors. A similar tendency for a slight reduction of the lateral incisors is also seen in Hex. sivalensis and Hex. imagunculus, and is found to an even greater extent in Hex. protamphibius turkanensis and Hex. coryndonae (Table VII). The latter species is unusual, however, in having 12 smaller than 13, Hexaprotodon karumensis from the upper Burgi Member, Hex. aethiopicus, Hip. gorgops, and Hip. amphibius are further derived in having a tetraprotodont arrangement of the lower incisors, in which

Hippopotamidae 155

Table vn. Comparison of Lower Incisors in Fossil and Extant Hippopotamids

Species

Hexaprotodon harvardi Lothagam & Lukeino Manonga Valley Kanapoi

Hexaprotodon sivalensis sivalensis

Trilobophorus afarensis Hexaprotodon imagunculus

Hexaprotodon protamphibius turkanensis

Hexaprotodon coryndonae Hexaprotodon karumensis

Koobi Fora (Upper Burgi) Hexaprotodon aethiopicus Hippopotamus gorgops Hippopotamus amphibius

Choeropsis liberiensis

Number of incisors

Hexaprotodont Hexaprotodont Hexaprotodont

Hexaprotodont

Hexaprotodont Hexaprotodont

Hexaprotodont

Hexaprotodont

Tetraprotodont Tetraprotodont Tetraprotodont Tetraprotodont

Diprotodont

Relative proportions of incisors

11- 3 subequal 11_ 3 subequal? 11 larger. 12-3 subequal (n = 1.

100:64:77)a 11 slightly larger than 13, 12 smaller

(n = 6. 100:73:88)b

11- 3 subequal ?II slightly larger than 12, 13 smaller

still (n = 3. 100:85:71)C 11 much larger. 12-3 subequal (n = 1.

100:50:51) 11 much larger than 13 , 12 smaller still

11 much larger than 12 (n = 4. 100:64) 11 much larger than 12 (n = 3. 100:66) 11 much larger than 12 (n = 2. 100:67) 11 much larger than 12 (n = 27. 100:63d;

n = 28. 100:5ge)

a Numbers in parentheses represent the relative size (maximum mesiodistal diameters) of the incisors. The largest incisor is equal to 100. They are average values arranged in sequence starting with I1. n is the sample size.

b From measurements presented by Hooijer (1950) the following comparative data can be calculated for Hex. sivalensis subspp.-sivalensis (100:79:93). namadicus (93:69:100). palaeindicus (89:36:100). sivajavanicus (100:93:70). koenigswaldi (100:78:86). and soloensis (100:80:92).

CData from Cooke and Coryndon (1970). Erdbrink and Krommenhoek (1975). d Data from Hooijer (1950). "Data from Pavlakis (1987).

the lateral incisor is only about two thirds of the diameter of the central incisor (Harris. 1991). Later samples of Hex. karumensis from the upper KBS Member are diprotodont (Harris, 1991), and in this respect they resemble the modern pygmy hippopotamus, Choeropsis liberiensis.

The lower incisors from the Manonga Valley are, therefore, similar to those of Hex. harvardi from Lothagam in their general morphology, and probably also in being subequal in size. In addition, they have the tip-to-tip wear pattern that is typical of Hexaprotodon.

4.3.2. Canines

The lower canine is bilaterally compressed and strongly backwardly recur­ved. It is D-shaped in cross section with a flattened or slightly concave medial face and a convex lateral face. The posterolateral surface of the crown is marked by a shallow longitudinal groove. Enamel is distributed evenly on the medial, anterior, and lateral sides, but is lacking from the posterior face. Apart from fine longitudinal striae, the enamel is perfectly smooth. Wear is concentrated on the enamel-free posterior portion of the tooth.

156 Chapter 6

The canines from the Manonga Valley are strongly bilaterally compressed, with a mean breadth-length index of 59 (n = 6; range = 53-65). This is a distinctive feature that they share with the samples from Lothagam and Lukeino (n = 9; mean index = 63; range = 57-76). Although other species of hippopo­tamids, both fossil and extant, exhibit a wide range of variation, the canines tend to be less compressed than those of Hex. harvardi, with mean values for this index of 65-67.

As in the upper canines, an index of relative lower canine size (maximum canine diameter x 100/mesiodistal length of M2) provides a useful basis for comparison. In the two extant species of hippopotami it is common for aged adults to have canines with a maximum diameter that exceeds that of the mesiodistal length of M2• The highest indices recorded in Hip. amphibius and C.liberiensis were 143 and 133, respectively. Comparative data on Hex. harvardi and other fossil hippopotamids indicate that they had canines no larger than those of the modern species, with a maximum index less than 130 (Le., Hex. harvardi, 110; Hex. sivaiensis, 120; Hex. karumensis, 128).

The lower canines from the Manonga Valley are typical of Hexaprotodon in having a smooth or finely striated enamel surface. They are comparable to Hex. harvardi from Lothagam and Lukeino in their size, general structure, and degree of bilateral compression.

4.3.3. Premolars

P1 is not represented in the collections from the Manonga Valley. P2 and P3

in Hexaprotodon are morphologically very similar, and apart from size, it is difficult to distinguish isolated teeth (Fig. 1). All four of the specimens that can be identified as either P2 or P3 probably represent P3• The single main cusp is tall, and distally recurved. The crown is mesiodistally long and narrow, and buccolingually slightly waisted in the mesial moiety of the crown. The mesial, distal, and distolingual crests are pustulate. A narrow but well-developed cingulum forms an almost continuous rim around the base of the crown. Perched on the distolingual margin of the crown is a prominent accessory cusp. The tooth bears two short roots. P 4 is represented by a single fragmentary specimen (WM 855/92). It is a molariform tooth, with an elevated main cusp, a prominent shelflike cingulum, and a distinct talonid with weakly developed pustules. The lingual portion of the crown is not preserved, so the development of the accessory cuspule, which is usually prominent in Hex. harvardi, is not known.

The lower premolars from the Manonga Valley are similar to Hex. harvardi from Lothagam and Lukeino in being relatively large in size, with well-devel­oped accessory cuspules, and a pustulate enamel surface. In terms of their occlusal areas (mesiodistal length x buccolingual breadth), the lower premolars all fall within the 95 % confidence limits for the combined sample from Lotha­gam and Lukeino (Table V). As discussed above, the relatively large size of the premolars is a distinctive characteristic of Hex. harvardi (Table VI).

Hippopotamidae 157

4.3.4. Molars

The lower molars are relatively long and narrow, with a slight waisting midway along the length of the crown (Fig. 1). The cusps are quite low and conical. As in the upper molars, each cusp has a sub-triangular-shaped exposure of dentine with slight wear, but this becomes a simple trefoil shape in later stages of wear. The mesial pair of cusps is larger and more elevated than the distal pair. The metaconid and hypoconid are linked by a low, rounded crest that passes obliquely across the talonid basin. Occasionally, a distinct metaconulid is present. On Ml and M2, a short crest originates from the hypoconid and passes distally, to give rise to a small hypoconulid on the cingular shelf. The hypo­conulid on Ma is low, but relatively large, and it is bordered buccally and lingually by well-developed accessory cuspules, the ectostylid and endostylid, respectively. The buccal and lingual cingula are absent, or are restricted to small conical tubercles or narrow shelves located at the base of the crown between the main cusps. The mesial and distal cingula are elevated and well developed. Morphologically, the lower molars from the Manonga Valley are indistinguish­able from those of Hex. harvardi from Lothagam and Lukeino.

The hypsodonty index (height of the crown x 100/buccolingual breadth of the crown) for the only complete and relatively unworn lower molars (M 44685 and WM 198/94) is 90 and 104, respectively. These teeth are relatively brachyo­dont compared with the lower molars from Lothagam and Lukeino, which have a range for this index of 92-113, and they also fall at the lower end of the general range for other species of Hexaprotodon from East Africa (89-127). However, two partially preserved Mas from the Manonga Valley are higher crowned, and their estimated hypsodonty index (111 and 113) coincides well with the upper limits for Hex. harvardi. In comparison with extant hippopotamids, the lower molars of Hex. harvardi are more similar to those of Choeropsis-with a maxi­mum hypsodonty index of 116-than they are to the distinctly high-crowned teeth of Hip. amphibius, which have a maximum index of 139.

In Hex. harvardi, the lower molars increase in size (mesiodistal length x buccolingual breadth) posteriorly, corresponding to the following ratio: 58:79:100. Similar size differentials between the lower molars are found in Hex. protamphibius (59:82:100) and in Hip. amphibius (60:76:100). Other hippopo­tamids, however, exhibit a more marked increase in size from Ml to Ma, as follows: Hex. sivalensis (43:66:100); Hex. karumensis (51:70:100); Hex. aethiopicus (54:74:100); Hex. imagunculus (58:66:100); Hip. gorgops (46:70:100); and Choeropsis liberiensis (53:83:100).

The lower molars from the Manonga Valley are comparable in size to those from Lothagam and Lukeino. In fact, based on their occlusal area (mesiodistal length x buccolingual breadth), all but one fall within the 95% confidence limits for the combined sample of lower molars from Lothagam and Lukeino (Table V). The exception is an Ma from Kiloleli 2, which is only slightly larger than the largest specimen from Lothagam. In comparison, the molars of Hex. harvardi are similar in size to those of Hex. sivalensis; smaller than those of Hex. karumensis, Hip. amphibius, and especially Hip. gorgops; slightly larger than those of Hex.

158 Chapter 6

protamphibius; and significantly larger than those of Hex. imagunculus, Hex. aethiopicus, and Choeropsis liberiensis.

In summary, the lower molars from the Manonga Valley are consistent in size and morphology with those of Hex. harvardi from Lothagam and Lukeino. They are distinguished from those of other species of Hexaprotodon from East Africa by the following combination of features: (1) larger size (except for those of Hex. karumensis, which tend to be larger still); (2) crowns more brachyodont and slightly narrower; and (3) less pronounced size differential between the molars (except for Hex. protamphibius).

4.3.5. Deciduous Dentition

The lower deciduous dentition is represented by four isolated teeth. The dP2

(M44688) is a narrow, elongated tooth, with a slight buccolingual waisting midway along its length. The single main cusp is tall. Originating from its apex are a low and rounded mesial crest, that bifurcates toward its base, and a pustulate distal crest. Basally, there is a strong cingulum that entirely encircles the crown. The dP3 is represented by a single incomplete and worn specimen (WM 656/90), which consists of the distal portion of the crown only. Mesially, the crown was relatively narrow, and it was dominated by a single tall cusp. The talonid basin has two well-developed cusps on its distal margin that are arranged in a transverse pair. A narrow but well-defined cingulum is continuous around the base ofthe distal moiety of the crown. The dP 4 is a low-crowned, molariform tooth, with a well-developed cingulum. The lower deciduous premolars are morphologically very similar to the corresponding teeth from Lothagam, but they tend to be slightly smaller in size.

5. Postcranial Material

In this section I present an account of the postcranial remains from the Manonga Valley. The intention is not to provide a comprehensive description of individual elements, but rather to highlight the structural differences that dis­tinguish the postcranium of Hex. harvardi from those of extant hippopotamids. Several authors have previously drawn attention to the importance of postcranial characteristics for differentiating Hexaprotodon and Choeropsis from Hippo­potamus (e.g., Hopwood, 1926; Arambourg, 1947; Cooke and Coryndon, 1970; Coryndon, 1977, 1978a,b; Geze, 1985; Pavlakis, 1990; Harris, 1991). The aim of this present analysis is to define these differences more precisely, and to use them as a basis for making general inferences about the possible locomotor behavior and habitat preferences of Hex. harvardi.

5.1. Vertebrae

Five isolated vertebrae have been recovered from the Manonga Valley, all from Tinde West (Table VIII). WM 010/90 consists of the centrum and posterior portion of the neural arch of an axis vertebra. It is comparable in size to that of female

Hi ppopotamidae 159

Table VIII. List of Hippopotamid Vertebrae and Limb Bones from the Manonga Valley Tinde

M44697 M44698 M44699 M44725 M44727

Tinde West WM 001/90 WM002/90 WM 010/90 WM 208/90 WM 222/90 WM 352/90 WM475/90 WM 574/90 WM 679/90 WM 666/92

Tinde East

Right distal humerus, lacking portion of distal articulation. Right distal tibia. Right patella, lacking inferior tubercle and medial margin. Distal radial epiphysis. Proximal radio-ulna, fragmentary.

Left distal tibia. Left distal tibia. Axis vertebra, centrum and caudal portion of neural arch. Right glenoid of scapula. Thoracic vertebra, almost complete. Caudal vertebra, probably caudal 3 or 4. Right patella, lacking inferior tubercle. Thoracic vertebra, portion of centrum and neural arch only. Right distal fibula. Lumbar vertebra, almost complete.

WM 316/90 & 456/90 Left distal humerus, two conjoining pieces. WM 317/90 Left distal tibia. WM 321/90 WM 557/92 WM 081/94

Kiloleli 2 WM 1286/92 WM 837/94

Inolelo 1 WM 140/94

Inolelo 3 WM 656/94

Beredi South 1 WM 1517/92

Ngofila 4 WM 1475/92 WM 1476/92 WM 1477/92

Right proximal tibia, medial portion only. Left distal humerus, ulnar trochlear only. Caudal vertebra.

Right distal tibia. Left distal tibia.

Right distal tibia.

Left distal radio-ulna.

Right distal tibia.

Right humerus, shaft and distal end only. Left distal femur. Right proximal radio-ulna, fragmentary.

individuals of Hip. amphibius, and is generally similar in morphology. However, it differs in the following respects: the posterior articular facets are situated closer together, suggesting a narrower neural arch dorsally; the centrum is anteropos­teriorly relatively shorter and dorsoventrally more compressed; and the dens is relatively more stout.

Two thoracic vertebrae are represented in the collections. WM 222/90 is an almost complete vertebra (lacking the transverse processes) from the posterior end of the thorax (probably Tll-13). WM 574/90 is more fragmentary, consisting of a portion of the centrum and neural arch from a midthoracic vertebra (probably T8-10). Both vertebrae are similar in most respects to Hip. amphibius, but they

160 Chapter 6

differ in having a mediolaterally broader, anteroposteriorly shorter, and dor­soventrally shallower centrum, and a more cranially oriented anterior costal facet.

WM 666/92 and WM 684/94 are relatively complete lumbar vertebrae, al­though their transverse and spinous processes have been lost. As in the thoracic vertebrae, the centrum is anteroposteriorly relatively much shorter than it is in Hip. amphibius. The proportions of the thoracolumbar vertebrae indicate that Hex. harvardi, like C. liberiensis, may have had a relatively shorter trunk than Hippopotamus.

WM 352/90 consists of a weathered and abraded 3rd or 4th caudal vertebra. WM 081/94 is a more fragmentary caudal vertebra, consisting of the centrum only. Compared with the corresponding vertebrae in modern Hippopotamus the fossils have relatively stouter transverse processes and a shorter centrum.

5.2. Pectoral Girdle and Forelimb

The forelimb is represented by several partial scapulae and a number of fragmentary limb bones (Table VIII). In addition, a sizable sample of carpals, metacarpals, and manual phalanges is also known, and these are discussed in section 5.4. The majority of these specimens comes from Tinde, but smaller collections have also been recovered from Inolelo, Ngofila, and Kiloleli (Tables IX and XI).

5.2.1. Scapula

Three scapula fragments have been identified as hippopotamid. WM 208/90 and WM 251/94 represent the distal portion ofthe scapula preserving the glenoid fossa, the coracoid process, and the base of the spine. WM 374/90 is more fragmentary, consisting of a portion of the glenoid and scapular neck only. They are comparable in size to those of female Hip. amphibius, and they are generally similar in morphology. The fossils differ from Hip. amphibius and C.liberiensis, however, in several respects. The coracoid process is shorter and stouter (similar in this respect to Choeropsis), the base of the scapular spine is relatively more robust, with the latter extending distally as far as the lateral margin of the glenoid, and the neck of the scapula is somewhat thicker. The glenoid cavity is mediolat­erally expanded, and much more nearly approaches a subcircular outline, rather than an oval. The breadth-length index of the glenoid articular surface in WM 208/90 (90.8) and wM 251/94 (86.5) corresponds with the upper limit of the range for extant hippopotamids (75.2-92.5).

5.2.2. Humerus

Only the shaft and the distal end of the humerus is known, being represented by three specimens from Tinde and a single specimen from Ngofila 4. They are all comparable in size to humeri of Hip. amphibius, but they differ in a number of respects. The distal articulation for the radio-ulna, for example, differs in being more strongly spooled, with a more pronounced median keel, and a more angular

Hippopotamidae 161

or raised lateral margin. The olecranon fossa is deep, proximo distally high, but mediolaterally quite narrow. The breadth-height index of the olecranon fossa in WM 1475/92 is 92.9. In Hip. amphibius the olecranon fossa tends to be relatively broader and more triangular in shape, rather than elliptical. The medial epicon­dyle in the fossils is more stoutly developed than in Hip. amphibius, presumably for the attachment of more powerful carpal and digital flexors. On the other hand, the lateral epicondyle and the lateral supracondylar ridge, to which the digital and carpal extensors attach, tend to be more weakly developed. In these respects the fossil humeri correspond more closely to the pattern seen in C. liberiensis.

5.2.3. Radius and Ulna

A right proximal radio-ulna, WM 1477/92 from Ngofila 4, consists of a portion of the shaft of the ulna, approximately 18 cm long, lacking most of the sigmoid notch and olecranon, as well as most of the proximal end of the radius. The radius is completely ankylosed to the ulna, except for a large, elliptical interosseus foramen that perforates between the two bones proximally for the passage of the posterior interosseus nerve and artery. The posteromedial aperture of the fora­men opens into a broad groove distally that eventually blends in smoothly with the general surface of the shaft. Similarly, the anterolateral aperture also opens into a shallow groove for the anterior interosseus artery and nerve. A similar pattern is typically found in young adults of modern hippopotami, although with increasing age and continued bone deposition, the groove develops into a well-defined gutter or partially enclosed canal that runs the entire length of the shaft.

The proximal articular surface of the ulna has a more pronounced median keel than in Hip. amphibius, and this accords well with the strong degree of spooling of the trochlea seen in the distal humerus. An additional proximal ulna fragment of a subadult individual is known from Tinde (M 44727). It preserves the olecranon process, lacking the unfused epiphysis, and the proximal half of the sigmoid notch. The olecranon process is longer and more robust than in Hip. amphibius, and it is more strongly posteriorly tilted. The proximal portion ofthe sigmoid notch is more strongly convex mediolaterally than in Hip. amphibius, and this is consistent with the deep and relatively narrow olecranon tunnel in the fossil humeri. The distal radius is represented by an abraded and unfused epiphysis (M 44725) and a partial radio-ulna (WM 656/94). They are consistent in morphology with both Hip. amphibius and C. liberiensis.

5.3. Hindlimb

The hindlimb is represented by a distal femur from Ngofila 4, several distal tibiae, a distal fibula and two patellae from Tinde, and distal tibiae from Beredi South, Kiloleli, and Inolelo (Table VIII). The material is similar in overall size to the corresponding bones of extant Hip. amphibius.

162 Chapter 6

5.3.1. Femur

The left distal femur from Ngofila 4 (WM 1476/92) is weathered and abraded. The main differences distinguishing the fossil from the femora of Hip. amphibius are that the patellar groove is deep, with more strongly keeled medial and lateral margins, the two condyles are anteroposteriorly longer and mediolaterally narrower, the intercondylar notch is relatively wider, and the scars for the collateral ligaments are more pronounced. These features, also characteristic of C.liberiensis, are functionally associated with increasing the stability at the knee joint during rapid movements in the parasagittal plane, and are presumably adaptations associated with fast-running, cursoriallocomotion.

5.3.2. Patella

Two patellae are known from the Manonga Valley, but unfortunately, both specimens are incomplete. WM 475/90 lacks only the apex for attachment of the patellar ligament, whereas M44699 is more fragmentary, lacking the apex and medial wing of the patella. The fossils are comparable in size to the patellae of female Hip. amphibius, but they are distinctive in being much more robust. In WM 475/90 the anteroposterior thickness of the bone is 80.3% of the mediolat­eral breadth, which is much more similar to other species of Hexaprotodon and to C.liberiensis (68.1-77.2%) than to Hip. amphibius (51.9-62.2%). In addition, the fossil patellae are similar to those of C. liberiensis, and differ from those of Hip. amphibius, in having a shorter and less curved medial wing, a feature linked with a better-developed medial keel on the patellar groove of the distal femur.

5.3.3. TIbia

A portion of a right proximal tibia is known from Tinde (WM 321/90). Some differences in the configuration and proportions of the articular surfaces for the distal femur distinguish the fossil from Hip. amphibius, but these are relatively minor and probably do not reflect important functional differences. The distal tibia, on the other hand, has a suite of functionally significant traits, consistent with those seen in the astragalus (see below), that serve to distinguish Hex. harvardi from Hip. amphibius. These are as follows: the distal end of the tibia is mediolaterally broader and more rectangular in outline; the styloid process on the posteromedial margin is higher and more conical; the midline keel is more elevated, and it is bordered medially and laterally by deeper articular grooves; and the pit for the medial ligament is deep, but not so extensive. In all of these respects the fossils resemble C. liberiensis. The shape and configuration of the distal articular surface of the tibia in Hex. harvardi is functionally associated with increased stability at the astragalocrural joint, allowing more rapid move­ments of the ankle joint in flexion and extension.

5.3.4. Fibula

The fibula is represented by a single well-preserved specimen, WM 679/90. In comparison with the distal fibula of Hip. amphibius it is a relatively stout bone, and the articular surface for the astragalus is more extensive and more

Hippopotamidae 163

obliquely oriented (its long axis is oriented at an angle of 48° to the long axis of the shaft). It is, however, comparable in morphology to the fibula in C.liberiensis.

5.4. Manus and Pes

5.4.1. Carpals

Examples of all elements ofthe carpus, with the exception ofthe trapezium, are known from the Manonga Valley (Table IX). The carpals are comparable in general size to those of Hip. amphibius (Fig. 2).

Table IX. List of Hippopotamid Carpals and Tarsals from the Manonga Valley

TInde Tarsals M 44694 M 44695 M 44696 M 44700 M 44726

Carpals M44721 M 44722 M 44723 M44724

Tinde West Tarsals WM 003/90 WM 004/90 WM 011/90 WM 028/90 WM 204/90 WM 205/90 WM 248/90 WM 259/90 WM 270/90 WM 304/90 WM 426/90 WM467/90 WM 685/90 WM 092/92 WM 118/92 WM 119/92 WM 121/92 WM 272/92 WM 351/92 WM 544/92 WM 615/92 WM 689/92 WM 707/92

Left astragalus. Left astragalus, lacking distal end. Right astragalus, missing proximolateral corner. Left navicular, fragmentary. Left calcaneum, proximal portion only.

Right cuneiform. Right cuneiform. Pisiform. Left unciform.

Left astragalus. Right astragalus. Left astragalus, slightly abraded, some superficial cracking. Right lateral cuneiform. Left calcaneum, proximal portion of bone only. Left astragalus. Left astragalus, lateral portion only. Left calcaneum, lacking distal end of bone. Calcaneum, heel process only. Right calcaneum, proximal portion only. Left astragalus. Left calcaneum, proximal end of heel process only. Left calcaneum, proximal portion only. Right cuboid. Right lateral cuneiform. Right navicular. Left navicular, fragmentary. Right lateral cuneiform. Right calcaneum, distal end only. Calcaneum, heel process only. Left navicular. Right navicular. Left astragalus.

(continued)

164

Carpals WM013/90 WM019/90. WM254/90 WM 287/90 WM 122/92 WM 263/92 WM299/92 WM470/92 WM 614/92 WM616/92 WM 709/92

Tinde East Tarsals WM 310/90 WM 039/94

Carpals WM 366/90 WM452/90 WM465/90 WM067/94 WM080/94

Kiloleli 2 Tarsals WM 857/92 WM 1242/92 WM385/94

Carpals WM 727/90 WM383/94 WM 384/94 WM386/94 WM 387/94

Table IX. (Continued)

Left scaphoid. Left trapezoid. Left uncifonn. Left cuneifonn. Left uncifonn. Right lunar. Left uncifonn. Left trapezoid. Left lunar. Right magnum. Left cuneifonn.

Left astragalus, lacking proximolateral margin. Left navicular.

Left uncifonn. Right lunar, fragmentary. Right lunar, fragmentary. Right scaphoid. Right cuneifonn, fragmentary.

Right astragalus. Left cuboid. Right cuboid.

Right cuneifonn. Left uncifonn. Right magnum. Left lunar. Left magnum, fragmentary.

Chapter 6

Two scaphoids are known. WM 067/94 is a complete and well-preserved right scaphoid. WM 013/90 is slightly damaged, with the trapezoid facet entirely lacking, and the articular surface for the magnum incompletely preserved. The bone is morphologically similar to that of Hippopotamus. However, it is dor­soventrally slightly more compressed. In addition, the articular surface for the radius is narrower, dorsoventrally more strongly concave, and mediolaterally less convex.

There are three almost complete and two partial lunars from the Manonga Valley (Fig. 2). The fossils are relatively more robust than those of modern Hippopotamus. The radial articular surface is similar in morphology to that in Hippopotamus. However, the articular surfaces for the unciform and magnum are relatively broader, being closer to a square in outline, rather than subrectan­gular, with distinct mediolateral waisting. In addition, the articular surface is

Hi ppopotamidae 165

---d

FIGURE 2. Comparison of carpals of Hex. hazvardi (left) with those of Hip. amphibius (right). (a) WM 263/92, right lunar, distal view; (b) WM 263/92, right lunar, lateral view; (c) WM 263/92, right lunar. proximal view; (d) WM 287/90, left cuneiform, lateral view; (e) WM 122/92, left unciform, distal view; (f) WM 122/92, left unciform, proximal view. Scale bar = 50 mm.

smoothly convex in the mediolateral plane, and it lacks the distinct keel that separates the two facets seen in Hip. amphibius. Furthermore, the V-shaped beaks that occur on the dorsal and ventral margins of the distal lunar surface in Hip. amphibius, associated with this keel, are less distinct and more rounded in the fossils.

The six fossil cuneiforms from the Manonga Valley are similar in size to those of Hip. amphibius, and they conform closely in morphology (Fig. 2). The main differences appear to be that the fossils are proximo distally longer in relation to the dorsoventral height of the bone, and that the articular facet for the unciform is taller and somewhat narrower.

The pisiform, M 44723, is a stout, cylindrical bone that recurves medially toward its apex. In its general proportions it is comparable to the corresponding bones of extant hippopotamids. The articular facets for the ulna and cuneiform are most similar in their configuration to those in C. liberiensis. In Hip. am­phibius. the cuneiform facet tends to be proximo distally narrower.

The unciform is represented by six specimens (Fig. 2). All are relatively complete, except for WM 366/90 and WM 383/94, which both lack the ventral process. They differ from the unciforms of Hip. amphibius and C. liberiensis in a number of respects: (1) the bone is dorsoventrally higher, mediolaterally narrower, and proximo distally thicker; (2) the styloid process is distinctly longer and more strongly distally recurved; (3) the facet for metacarpal III is relatively much larger, and is less clearly demarcated from the facet for the magnum; (4)

166 Chapter 6

the facet for the magnum has a convex dorsal margin (in Hip. amphibius it is V-shaped, which interlocks with a corresponding depression in the magnum); (5) the facets for metacarpals IV and V are dorsoventrally higher and narrower; (6) the facets for metacarpal IV and V are separated inferiorly by a narrow interosseus sulcus, which terminates about one-third down from the dorsal margin at a deep ligamentous pit (in Hip. amphibius the two facets are contigu­ous, and a shallow ligamentous pit occurs more inferiorly); (7) the facet for metacarpal IV is generally dorsoventrally higher or sub equal in height to the facet for metacarparpal V, whereas in Hip. amphibius the reverse is typically the case; (8) the facet for the semilunar is more distinctly sellar, being mediolaterally more concave, with a tighter arc of curvature dorsoventrally, and extending further onto the dorsal aspect of the bone; (9) the facet for the cuneiform also exhibits a more strongly developed saddle; and (10) the lateral wing of the bone is proximo distally thicker, and this leads to a wider separation between the facets for metacarpal V and the cuneiform (in Hip. amphibius the two facets almost touch).

Three partially preserved magnums are known (WM 616/92, WM 384/94, WM 387/94), each lacking the apex of the styloid process. As is generally the case in the other carpals, the magnum is mediolaterally relatively narrower than in Hip. amphibius and C. liberiensis. Medially, the dorsal facet for metacarpal II is smaller, and it has a narrower contact with the facet for the trapezoid. The articular surface for metacarpal III is mediolaterally narrower. The facet for the unciform is reniform, concave, and quite deeply excavated, but it lacks the V-shaped depression commonly seen in extant hippopotamids that allows the magnum and unciform to interlock.

The trapezoid is represented by two specimens. WM 019/90 is complete; WM 470/90 is somewhat weathered and abraded. They differ from those of Hip. amphibius in the following ways: (1) the bone is relatively mediolaterally narrower; (2) the facet for the scaphoid is narrower and mediolaterally more strongly concave; (3) the distal articular surface for metacarpal II is narrower, flatter, and more elliptical, rather than triangular in shape; and (4) there is no development of a beak on the dorsal aspect of the bone to interlock with the magnum.

The carpus of Hex. haTVardi, therefore, differs in a number of respects from Hip. amphibius, and is much more similar to C. liberiensis. Perhaps the most striking difference, at least from a functional perspective, is that the wrist in Hex. haTVardi is mediolaterally relatively narrower and more compact than in Hippo­potamus, an adaptation in ungulates associated with a superior capability for fast running.

5.4.2. Tarsals

The astragali of Hex. haTVardi are well represented in the collections from the Manonga Valley. They are similar in size or slightly smaller than those of Hip. amphibius, but they can be readily distinguished on the basis of a suite of morphological features (Fig. 3). These are as follows: (1) the fossil astragali are relatively narrower (Table X); (2) the saddle-shaped sustentacular facet for the

Hippopotamidae 167

a

e f ---g h

FIGURE 3. Comparison of tarsals of Hex. harvardi (left) with those of Hip. amphibius (right). (a) WM 004/90, right astragalus, superior view; (b) WM 004/90, right astragalus, inferior view; (e) WM 269/90, left calcaneum, medial view; (d) WM 269/90, left calcaneum, superior view; (e) WM 119/92, right navicular, distal view; (f) WM 119/92, right navicular, proximal view; (g) WM 118/92, right lateral cuneiform, distal view; (h) WM 118/92, right lateral cuneiform, proximal view. Scale bars = 50 mm. Top scale bar for a-d, bottom scale bar for e-h.

calcaneus is less strongly contoured; (3) the sustentacular facet is bordered distolaterally, medially, and proximo medially by deep and well-defined pits for the attachment of interosseus ligaments; similar depressions occur in the astra­gali of Hip. amphibius, but they tend to be even more strongly developed; (4) the naviculocuboid facet is similar in shape and proportions to that in modern Hippopotamus, except that the midline keel is more rounded in the fossil astragali, and the cuboid facet describes a tighter arc of curvature; (5) the trochlear surface for articulation with the distal tibia is relatively narrower, more deeply grooved, and with steeper medial and lateral walls; (6) the lateral margin of the tibial trochlea projects much further distally, and exhibits a smaller arc of curvature; (7) the medial margin of the tibial trochlea extends further round onto the inferior surface ofthe bone; (8) the naviculocuboid facet is much more widely separated from the lateral trochlear margin, and the roughened interosseus sulcus separating them is more deeply excavated; (9) the cuboid facet and the distal calcaneal facet meet at a sharper angle (the mean angle is 1100 in the fossils and 1230 in Hip. amphibius) , and the cuboid facet is more laterally facing; (10) the fibula facet is more extensive and slightly convex; (11) the lateral calcaneal

168 Chapter 6

Table X. Breadth-Length Index of Astragalus in Fossil and Extant Hippopotamidsa

N Mean Range

Hexaprotodon harvardi (Lothagam. Lukeino. Mpesida)

Hexaprotodon harvardi (Manonga Valley)

Hexaprotodon sivalensis Hexaprotodon protamphibius Hexaprotodon aethiopicus Hexaprotodon karumensis Kenyapotamus cOIyndonae Hippopotamus gorgops Hippopotamus amphibius Choeropsis liberiensis

a Index = maximum length of astragalus x 100

maximum breadth of astragalus

22

6

10 6 6 7 1 1

10 5

70.6

71.0

72.0 75.9 73.6 72.6 72.1 84.1 82.9 72.4

62.3-76.8

68.2-73.3

65.7-78.8 71.8-77.6 65.5-78.1 69.9-76.5 72.1 84.1 76.5-92.3 70.0--73.9

facet is represented by a small elliptical or triangular depression; in Hip. amphibius it is a large L-shaped facet (Fig. 3). In these respects, and in their general morphology, the fossil astragali are closely comparable to those of C. liberiensis. In comparison with Hip. amphibius, the proportions of the astragalus and the configuration of the articular facets for the tibia and fibula in Hex. harvardi imply that there was a greater range of flexion at the astragalotibial joint, especially in plantarflexed positions, and increased stability for speed of action in flexion and extension.

There are no complete calcanei from the Manonga Valley. Most specimens comprise the tuber calcis only, but several specimens preserve portions of the sustentaculum tali, the proximal and distal astragalar facets, and the cuboid facet (Fig. 3). Together these specimens permit a fairly complete composite descrip­tion of the entire bone. The fossil calcanei are generally similar in size to female individuals of Hip. amphibius, but differ in a number of morphological details. For example, the heel process in the fossil calcanei is deeper and mediolaterally more slender. The proximal extremity of the heel process is more elliptical (rather than square) in outline, and it bears a deeper groove for a well-developed calcaneal tendon. Consistent with differences observed in the astagalus, the proximal calcaneal facet is flatter and more triangular in shape. The facet for the cuboid differs from that in modern Hippopotamus in being relatively shorter dorsoventrally, and in being mediolaterally slightly concave, rather than flat to slightly convex. In these respects the fossils conform closely to the morphologi­cal pattern in C. liberiensis.

Two cuboids (WM 092/92 and WM 385/94) are known from the Manonga Valley. The general construction of the bone is similar to that of both extant species of hippopotamus, but it differs from that of Hip. amphibius in being relative shorter in the proximodistal plane, in having a less strongly dorsoven-

Hippopotamidae 169

trally concave astragalar facet, narrower articular facets for metatarsals IV and V, and deeper pits for ligamentous attachments to the astragalus.

Four fairly complete and two partially preserved naviculars are known. They are similar in morphology to those of C. liberiensis, but differ from Hip. am­phibius in being mediolaterally narrower, having a narrower talar facet, with a wider arc of curvature and a more distinct dorsoventral keel, smaller and flatter articular surfaces for the cuneiforms, and a more extensive articular contact with the cuboid (Fig. 3).

Three almost complete lateral cuneiforms have been recovered from the Manonga Valley (Fig. 3). Apart from some minor differences in the shape of the navicular facet, the fossils are very similar in morphology to the lateral cunei­forms of C. liberiensis and Hip. amphibius. No middle and medial cuneiforms are known.

5.4.3. Metapodials

Metapodials are well represented in the collections from the Manonga Valley (see Table XI). They are readily distinguished from those of Hippopotamus in the following respects: (1) they are relatively longer and more slender; (2) the distal articular head is dorsoventrally shallower, with a more pronounced articular keel, and it tends to become increasingly narrower dorsally and does not extend as far onto the dorsal aspect of the bone; (3) on the dorsal surface of the shaft, just proximal to the distal articular surface, there is commonly a deep depression for the attachment of the capsule (this depression tends to be absent

Table XI. List of Hippopotamid Metapodials and Phalanges from the Manonga Valley Metacarpal II

TInde East, WM 346/90 (proximal) Kiloleli 3, WM 789/90 Inolelo 3, WM 657/94

Metacarpal III TInde West, WM 005/90 (proximal) TInde East, WM 448/90 (proximal), WM 345/90 (proximal), WM 361/90 Inolelo 3, WM 658/94 (proximal)

Metacarpal IV TInde West, WM 027/90 (proximal) TInde East, WM 348/90 (proximal) Kiloleli 2. WM 840/92 (proximal) Inolelo 3, WM 658/94 (proximal)

Metacarpal V TInde West, WM 023/90 (proximal), WM 495/90, WM 576/92 (proximal) TInde East. WM 1843/92 (proximal)

Metatarsal II TInde, M 44707 (proximal) Kiloleli 2, WM 839/92

Metatarsal III TInde, M 44708 (proximal) Tmde West, WM 476/90 Kiloleli 2, WM 842/92 (proximal), WM 843/92

(continued)

170

Metatarsal IV Inolelo 3, WM 659/94 (proximal)

Metatarsal V

Table XI. (Continued)

nnde West, WM 249/90, WM 688/92, WM 563/92 (distal) Miscellaneous metapodials

nnde, M 44703 (distal), M 44705 (distal), M 44706 (proximal), M 44709 (distal)

Chapter 6

TInde West, WM 275/90 (distal), WM 482/90 (distal), WM 606/92 (distal), WM 543/92 (distal), WM 018/94 (distal), WM 913/94 (distal)

nnde East, WM 333/90 (distal), WM 460/90 (distal), WM 686/90 (distal) Kiloleli 2, WM 708/92 (distal), WM 844/92 (distal), WM 388/94 (distal), WM 389/94 (distal)

Proximal phalanges nnde, M 47701, M 47702 (distal) nnde West, WM 014/90 (fragment)' WM 016/90, WM 037/90 (proximal), WM 041/90 (frag­

ment), WM 421/90, WM 424/90 (fragment), WM 425/90, WM 477/90, WM 488/90, WM 491/92 (proximal), WM 609/92 (proximal), WM 610/92 (proximal), WM 677/92 (fragment), WM 710/92a (fragment), WM 146/92, WM 520/92, WM 019/94. WM 103/94. WM 916/94

nnde East, WM 341/90. WM 353/90 (fragment), WM 371/90, WM 453/90. WM 1844/92. WM 1845/92 (fragment). WM 1848/92 (proximal epiphysis)

Kiloleli 2, WM 849/92, WM 1285/92 (distal) Kiloleli 3. WM 746/90. WM 788/90. WM 791/90 Kiloleli 4. WM 524/94 (proximal) Shoshamagai 2, WM 1942/92 Inolelo 1. WM 1031/92 (proximal), WM 1117/92 Inolelo 3, WM 686/94 (proximal)

Middle phalanges nnde West, WM 050/90; WM 286/90. WM 468/90, WM 116/92 (proximal). WM 206/92. WM

400/92 (proximal). WM 607/92 (proximal). WM 608/92 (proximal), WM 710/92b (proximal). WM 611/92 (lacking proximal end). WM 612/92, WM 613/92. WM 774/92 (immature, lacking proximal epiphysis), WM 1896/92 (distal). WM 1999/92 (proximal), WM 123/94. WM 917/94 .

nnde East. M 329/90, WM 462/90. WM 1846/92. WM 040/94 (proximal), WM 060/94 Kiloleli 2. WM 848/92, WM 1284/92 (proximal), WM 391/94a. WM 391/94b Kiloleli 3. WM 946/92 Inolelo 1, WM 161/94

Distal phalanges nnde West. WM 102/94 Kiloleli 2. WM 1287/92 (distal end missing) Inolelo 2. WM 634/94

or very shallow in Hip. amphibius); and (4) the proximal articular surface tends to be narrower, but dorsoventrally deeper (Fig. 4). In most of these respects, the fossils correspond closely to metapodials of Choeropsis.

The most obvious feature that distinguishes the metapodials of Hex. harvardi from those of Hip. amphibius and C. liberiensis is that they are longer and more gracile. Comparative data on their relative robusticity show that Hex. harvardi has metapodials that are on average 16.5% less robust than Hip. amphibius and 15.9% less robust than C. liberiensis. However, relatively long and slender metapodials appears to be a general characteristic of Hexaprotodon (Table XII).

Another characteristic that purportedly distinguishes Hexaprotodon and Choeropsis from Hippopotamus is the occurrence of relatively shorter lateral

Hippopotamidae 171

Metacarpals

--. Metatarsals

II III v

FIGURE 4. Comparison of metapodials of Hex. harvardi (right) with those of Hip. amphibius (left). Top row - MC II, WM 789/90; MC III, WM 361/90; MC V. WM 495/90. Bottom row - MT II. 839/92; MT III, 843/92; MT V, 249/90. Scale bar = 50 mm. Note the greater length and slenderness of the fossil metapodials.

Table XII. Mean Robusticity of Metapodials in Fossil and Extant HippopotamidsQ

Metacarpals Metatarsals

II III IV V II III IV V Mean

Hex. harvardi 19.3 17.0 24 .9 24.1 19.8 18.4 21 .3 20.7 Hex. karumensis 20.7 18.5 17.7 25.1 26.8 20.2 20.7 25.1 21.9 Hip. gorgops 22.9 23.1 31.8 26.8 24.8 25.0 28.9 26.2 Hip. majorb 26.1 24.9 28.7 35.5 32.4 28.9 28.9 33.5 29.9 Hip. amphibius (n = 14) 23.1 20.0 23.0 29.6 28.7 22.8 23.1 28.0 24.8 C. liberiensis (n = 3) 25.7 19.4 25.3 25.2 28.3 24.2 23.0 26.0 24.6

a Robusticity of metapodials = "cross-sectional area of metapodial total length of metapodial

bCalculated from data published by Faure (1985).

172 Chapter 6

digits in comparison with the median digits (Coryndon, 1977; Geze, 1985; Harris, 1991). This observation is confirmed for the extant hippopotamids. Owing to the lack of associated pedal and manual material for most extinct species, it is not possible to calculate their relative ray length. Nevertheless, a good indication of relative size of the lateral rays can be obtained using data from the metapodials alone. In extant hippopotamids, for example, Choeropsis has much shorter lateral metapodials in relation to the length of the median metapodials than does Hip. amphibius (Table XIII). However, contrary to previous reports, comparative data for the fossil hippopotamids indicate that Hexaprotodon was more like Hip. amphibius, rather than Choeropsis, in having relatively long lateral rays. The isolated metapodials from the Manonga Valley provide additional confirmation that the lateral rays of Hex. harvardi were relatively long (Table XIII).

As noted above, a number of features of the distal metapodial serve to distinguish Hex. harvardi and C. liberiensis from Hip. amphibius. These differ­ences presumably relate to increasing the stability of the metapodial-phalangeal joint during flexion and extension, especially important in ungulates with relatively elongated digits. The narrowing of the head dorsally, and its limited extension onto the dorsal aspect of the bone, suggests that stability was optimized during flexed postures, and that the potential for hyperextension at the joint was rather limited in comparison with Hip. amphibius. As Yalden (1971) has noted, the effectiveness of the flexion hinge of the manus is improved by limiting the range of hyperextension that is possible at these joints. A number of sesamoid bones have been recovered from Tinde and Kiloleli, and these are identical to the paired sesamoids associated with the distal metapodials in extant hippopo­tamids.

Metacarpal II is represented by complete, but slightly weathered, specimens from Kiloleli 3 (WM 789/90) and Inolelo 3 (WM 657/94), as well as a partial specimen from Tinde East (WM 346/90). The proximal articular surfaces differ

Table XIII. Relative Ray Length in Hippopotamids Relative length of lateral raysa

Relative length of lateral metapodialsb

C. liberiensis (n = 4) Hip. amphibius (n = 4) Hex. harvardi Hex. karumensis Hex. aethiopicus Hip. gorgops Hip. majorC

Manus Pes

75.3 84.3

77.4 75.3

mean length of lateral rays (II and V) x 100 a Relative length of lateral rays = I th f d· d

mean eng 0 me Ian rays (Ill an IV)

Manus

67.6 79.8

77.6 74.7

80.1

b mean length of lateral metapodials (II and V) x 100 Relative length of lateral metapodials = --..,....::.------::-~--:----­

mean length of median metapodials (III and IV) CData from Faure (1985).

Pes

66.4 69.6 72.8 72.0 69.5 74.0 76.0

Hippopotamidae 173

from those in Hip. amphibius in that the facet for the trapezoid is narrower and less markedly concave mediolaterally, the facet for the magnum is narrower, and that for metacarpal III is more extensive. A complete metacarpal III is known from Tinde East (WM 361/90), and a proximal metacarpal III, preserving much of the shaft, is known from Inolelo 3 (WM 658/94). The articular surface for the unciform is relatively smaller and more triangular in shape by comparison with that in Hip. amphibius, and the facet for the magnum tends to be dorsoventrally deeper. Metacarpal IV is represented by four incomplete specimens, preserving the proximal portion of the bone only. The unciform facet is narrower and dorsoventrally deeper, and the facet for metacarpal III is more closely applied to the medial side of the bone than is typically seen in Hip. amphibius. Metacarpal V, known from a single complete specimen (WM 495/90) and several fragmentary specimens from Tinde, differs from Hip. amphibius in having a narrow proximal articular surface, a less prominent saddling of the unciform facet, a less distinct styloid process, and a relatively smaller facet for metacarpal IV.

From the pes, metatarsal II and III are represented by complete specimens, metatarsal IV is known from a proximal fragment only, and metatarsal V is represented by several fragmentary specimens. Apart from the general charac­teristics of the metapodials described above, the main difference that distin­guishes the metatarsals from those of Hip. amphibius is that the main proximal articular surfaces are mediolaterally narrower. This is consistent with the evi­dence from the tarsals that the pes of Hex. harvardi was relatively narrow.

5.4.4. Phalanges A reasonable sample of isolated phalanges has been recovered from the

Manonga Valley, but unfortunately, it has proved extremely difficult to sort them to their specific ray, or even whether they belong to pedal or manual digits. Nevertheless, it is possible to arrange them according to whether they belong to median or lateral digits (Table XI). Morphologically, they are closely similar to those of Choeropsis, but they differ in a number of respects from those of Hip. amphibius.

The proximal phalanges are relatively longer and more slender than those of Hip. amphibius. The index of robusticity of the proximal phalanges is signifi­cantly lower in Hex. harvardi than in either of the extant species, especially in Hip. amphibius (Table XIV). When combined with evidence from the metapo­dials, it is clear that Hex. harvardi had relatively elongated cheiridia, much more suited to fast-moving cursoriality. The distal articular surface of the proximal phalanges is dorsoventrally deeper, and it narrows dorsally more markedly than in Hip. amphibius. This arrangement serves to maximize the stability of the interphalangeal joint during flexion and extension, with greatest support being given when the middle phalanx is in a semiflexed position. By contrast, the broader articular surface in Hip. amphibius, especially on the dorsal aspect of the joint, permits greater stability when the middle phalanx is fully extended or even hyperextended. The proximal articular surface for the metapodial tends to be relatively narrower and more concave mediolaterally, the sagittal articular groove is deeper, and the medial, dorsal, and lateral margins are more sharply

174 Chapter 6

Table XIV. Relative Robusticity of the Phalanges in Hexaprotodon harvardi and Extant HippopotamidsQ

Hex. harvardi Hip. amphibius C. liberiensis

N Mean Range N Mean Range N Mean Range Proximal phalanges,

lateral digits (II & V) 5 35.5 32.7-38.3 9 45.2 43.3--47.9 6 40.5 38.6--44.2 Proximal phalanges,

median digits (Ill & IV) 8 37.7 33.9--40.6 9 42.3 40.7--44.4 7 40.2 37.7--41.9 Middle phalanges,

lateral digits (II & V) 9 60.0 51.4-71.6 8 61.4 58.4-65.8 6 62.1 60.2-64.1 Middle phalanges,

median digits (Ill & IV) 4 68.1 57.8-76.8 8 60.7 56.9-65.6 6 59.3 57.8-62.2

a Index = ";cross-sectional area of midshaft of phalanx total length of phalanx

defined, with a much better defined dorsal beak. These differences are function­ally associated with improved stability at the metapodial-phalangeal joint, especially during rapid flexion and extension. Similarly, the middle phalanges can be distinguished from those of Hip. amphibius in being slightly less robust, with a proximal articular surface that shows adaptations for increased stability during parasagittal movements at the interphalangeal joint (Le., the facet is mediolaterally narrower, the midsagittal keel is more pronounced, and the dorsal and ventral beaks are more projecting). Terminal phalanges have only been recovered from Kiloleli 2 and Inolelo 2. They are comparable in size, proportions, and morphology to those from the median manual digits of Hip. amphibius.

An isolated phalanx from Tinde West (WM 520/92) is noticeably much smaller than all of the other phalanges from the Manonga Valley. The specimen is a complete proximal phalanx from a lateral ray, probably from the pes. As both epiphyses are intact and fully fused, it is clearly from an adult individual. In its linear dimensions (the length of the phalanx is 42.1 mm) it is on average only 66% the size of the next smallest proximal lateral phalanx (which, incidentally, is about the same order of magnitude as that between phalanges of Hip. am­phibius and C. liberiensis). Apart from its small size, however, the phalanx is similar in morphology and proportions to the other phalanges from Tinde. The diminutive size of WM 520/92 precludes it from assignment to Hex. hmvardi, and it establishes the presence in the Manonga Valley of a much rarer pygmy species of Hexaprotodon. In terms of its size, it is quite a bit smaller than phalanges attributed to Hex. aethiopicus, a pygmy hippopotamus from the Turkana basin (Harris et aI., 1988; Harris, 1991). It is pertinent here to note that a small species of Hexaprotodon, of uncertain taxonomic status, also appears to have coexisted with Hex. harvardi at Lothagam and Lukeino (Coryndon, 1978b; M. G. Leakey, pers. comm.; Harrison, unpublished). Based on craniodental evidence, however, it appears that this latter species was a little larger than Hex. aethiopicus, but was comparable in size to Hex. imagunculus from the Western Rift.

Hippopotamidae 175

5.5. Functional and Behavioral Implications of the Postcranium

The morphology of the forelimb of Hex. harvardi. especially the elbow joint. differs in a number of important respects from that of Hip. amphibius. and more closely resembles the pattern observed in C. liberiensis. In contrast to Hip. amphibius. the distal humeri of Hex. harvardi and C. liberiensis have a more strongly spooled trochlea. better development of the median keel. and a rela­tively deep and narrow olecranon fossa. These features are adaptations for increasing the stability of the elbow joint during rapid flexion and extension of the forearm. In addition. indications that the carpal and digital flexors may have been more strongly developed in Hex. harvardi are consistent with this func­tional interpretation. In ungulates the wrist joint acts as a "flexion hinge" in which the manus is folded during protraction of the forelimb to avoid dragging the digits (Yalden. 1971). Therefore. Hex. harvardi. with its combination of adaptations for rapid parasagittal movements of the forearm and its relatively long digits. would require well-developed digital and carpal flexors. Similar indications of adaptation for greater stability and an increased range of motion in the parasagittal plane are also found in the knee and astragalocrural joints. These adaptations. together with the more lightly built postcranial skeleton, longer and more slender limb bones. narrower and more close-packed carpus and tarsus, and longer metapodials and phalanges. suggest that Hexaprotodon was a more digitigrade. fast-moving animal, somewhat less specialized than Hippopotamus for an amphibious habit. The postcranium of Hex. harvardi is. however. very similar to that of C. liberiensis. remarkably so given the consider­able difference in estimated body size between the two species (Le .. 250-275 kg for C. liberiensis and -1000-2000 kg for Hex. harvardi).

6. Taxonomy and Phylogenetic Relationships

A comparison with other early hippopotamids is necessary to place the Manonga Valley material in its appropriate taxonomic. phylogenetic. and bio­chronological context. Following a review of its relationships at the generic level. a more detailed comparison of the material with early hippopotamids from East Africa. North Africa. and Eurasia is presented. Most of the observations dis­cussed below are based on studies of original materials. but in a few cases. where noted. my comments refer to published descriptions. illustrations. and measure­ments.

6.1. Generic Affinities of the Manonga Valley Hippopotamid

6.1.1. Hexaprotodon

It is evident from the preceding descriptions of the craniodental and postcra­nial material that the large species of hippopotamid from the Manonga Valley can be referred to the genus Hexaprotodon. A number of primitive traits serve to distinguish Hexaprotodon from Hippopotamus. These include: upper and

176 Chapter 6

lower incisors with tip-to-tip occlusion and relatively flat apical wear; upper incisors set in a shallow parabolic series; upper canines with a deep posterior groove; upper and lower canines with finely striated enamel; premolars rela­tively large; molars brachyodont with well-developed cingula, tapered lobes, and a triangular enamel wear pattern; a short muzzle and elongated neurocranium; orbits situated laterally, approximately midway along the length of the cranium; a tendency for the median suture of the premaxillae to be fused anteriorly; lacrimals small and separated from the nasal bones by a well-developed antor­bital process of the frontal bone; nasals expanded posteriorly; limb bones with joint surfaces that maximize stability for parasagittal movements; and elongated metapodials and phalanges (Coryndon, 1978a; Gaze, 1985; Harris, 1991).

6.1.2. Trilobophorus

In addition to Hexaprotodon and Hippopotamus, Geze (1985) recognized a third genus of Plio-Pleistocene hippopotamid from East Africa, which he for­mally named Trilobophorus. The taxon is known only from the late Pliocene sites of Geraru and Hadar (Hadar Formation) in the Awash Valley (Geze, 1985; Kalb & Mebrate, 1993). I have not had the opportunity to study the original material from Ethiopia, but the diagnosis presented by Gaze (1985), which mainly includes features of the skull and dentition typical of Hexaprotodon, provides insufficient grounds for the recognition of a separate genus. Based on the few unique characteristics listed by Geze (1985), the material is tentatively retained as Hex. afarensis, although further comparisons are needed in order to determine its distinctiveness from Hex. protamphibius turkanensis.

6.1.3. Kenyapotamus The earliest hippopotamids are known from the middle and late Miocene of

East Africa. Pickford (1983) has described two species-Kenyapotamus ternani, from the middle Miocene of Fort Ternan and Maboko Island (12-15 Ma), and Kenyapotamus coryndonae, from the late Miocene of Ngeringerowa, Ngorora Formation, Nakali, and Namarungule Formation (7-10 Ma)-that he includes in a separate subfamily of the Hippopotamidae, the Kenyapotaminae. In addition, Pickford (1990b) has described the remains of Kenyapotamus from the Beglia Formation in Tunisia, dated at 9-10 Ma, which are close in morphology to those of K. coryndonae. There is good evidence to support the inference that Kenyapo­tamus represents the sister taxon of all later hippopotamids (hippopotamids of "modern aspect"), which are grouped together in the Hippopotaminae. Kenyapo­tamines are distinguished from hippopotamines primarily by the retention of more primitive traits. Kenyapotamus, for example, is much smaller than Hex. harvardi (the occlusal areas of the cheek teeth of K. tern ani and K. coryndonae are only 24% and 38%, respectively, of those of Hex. harvardi) , and it has simpler, more bunodont cheek teeth, with occlusal details closer to the ancestral suiform morphotype (Pickford, 1983). It would seem that the kenyapotamines, which replaced the anthracotheres as the dominant group of large amphibious mammals at the end of the early Miocene, at least in the local East African setting, were themselves superceded by the hippopotamines during the late Miocene

Hippopotamidae 177

(6-7 Ma) (Coryndon, 1978a; Pickford, 1983). As the time interval between the last kenyapotamines and the appearance of the earliest hippopotamines is relatively slender (probably about 1 m.y.), and since the extent of the morpho­logical differences between the two groups it striking, it seems likely that Hexaprotodon was an immigrant into East Africa during the late Miocene. Pickford (1983) has put forward the intriguing suggestion that increased tectonic activity in the Rift Valley system during this period may have impeded rivers and led to the development of a mosaic of lakes and lakeside environments, which established, possibly for the first time, suitable niches for the hippopo­tamines to expand their geographic range into East Africa.

6.1.4. Choeropsis

Further support for the hypothesis that hippopotamines may have originated outside of East Africa comes from the biogeography and phylogenetic affinities of the living hippopotamids. The extant pygmy hippopotamus, Choeropsis liberiensis, from West Africa, appears to be the sister taxon of all other hippopo­tamines, both living and fossil. Apart from a few autapomorphies ofthe dentition and postcranium, C. liberiensis conforms closely to the inferred ancestral mor­photype of the Hippopotaminae, and in the structure of the face and neurocra­nium it would seem to be more primitive even than Hex. harvardi. I agree with Coryndon (1978a) and Pickford (1983), therefore, that the lineage leading to the modern pygmy hippopotamus diverged early, being derived from a common ancestor that was more conservative than any known fossil representative (except for Kenyapotamus) (see Fig. 5). This implies that the divergence of the Hippopotaminae occurred prior to their initial appearance in the fossil record in East Africa, and that, since the most primitive member of the subfamily, the extant pygmy hippopotamus, has a West African distribution, there is at least circumstantial evidence that the hippopotamines may have originated outside of East Africa.

Moreover, the inferred relationship between the living pygmy hippopotamus and the other hippopotamids discussed above has some important ramifications for the taxonomy ofthe group. Coryndon (1977) argued that Choeropsis liberien­sis should be included in the genus Hexaprotodon, and this viewpoint has generally been adopted by subsequent workers. However, the marked similarities between Hexaprotodon and Choeropsis are undoubtedly a consequence of the retention of primitive hippopotamine traits, and do not provide evidence of a particularly close phylogenetic relationship. The current evidence would favor, in fact, the interpretation that the modern pygmy hippopotamus is a somewhat specialized derivative of the sister taxon of all other hippopotamines. In this case a separate generic designation seems appropriate.

6.1.5. Hippopotamus

A number of derived characteristics of the dentition, skull, and postcranium serve to distinguish Hippopotamus from all other hippopotamines. These in­clude: tetraprotodont upper and lower incisors, with scissorlike occlusion; upper canines with a shallow posterior groove; upper and lower canines with

178

Ma

2

4

6

8

;: ",,£ ::J '"

- II> ::J ::J v .. . ~ ~ .)( .. ...

"'::c ::c_

'" ::J E '" "0 D­.. >­c ...

::.::

'" D-o

, '. ff lkl tIO

ci.. :t '

II> 'Vi c ... ·c ... ..D

Chapter 6

FIGURE 5. Phylogenetic relationships of African and Eurasian hippopotamids. The scheme is a provisional one, including only those taxa discussed in the text.

heavily grooved enamel; premolars small; molars elongated and more hypso­dont, with poorly developed cingula, and a trefoliate enamel wear pattern; a long muzzle and an abbreviated neurocranium; orbits tend to be elevated superiorly; the median suture of the premaxillae remains unfused anteriorly; lacrimal large, making contact with the nasal bones; nasals expanded posteriorly; limb bones, metapodials, and phalanges short and robust, and adapted for a more amphibious habit (Coryndon, 1978a; Geze, 1985; Harris, 1991).

These synapomorphies link Hip. amphibius with several fossil species from the Plio-Pleistocene of Africa (Le., Hip. kaisensis and Hip. gorgops), and define them as a clade distinct from all other hippopotamines. Hippopotamus was

Hippopotamidae 179

presumably derived from an advanced member of the Hexaprotodon lineage, probably some time during the later part of the Pliocene (see Fig. 5). However, if Hippopotamus does indeed share a more recent common ancestor with one of the more derived species of Hexaprotodon than it does to primitive hippopo­tamines (such as Hex. harvardl), then we are forced to concede that the genus Hexaprotodon, as currently construed, represents a paraphyletic taxon. Hexaprotodon may be easy to diagnose on the basis of an extensive suite of characters (see description above), but these features appear to be entirely symplesiomorphic. In this case, the taxon has no integrity as a clade; it merely represents a cluster of stem species of hippopotamines that lack certain derived features characteristic of Hippopotamus. Clearly, a more detailed assessment of the phylogenetic relationships is required before the taxonomy of hippopo­tamines adequately reflects their cladistic affinities. It may be that Pickford's usage of Hippopotamus to encompass Hexaprotodon, while retaining Choerop­sis as a distinct genus, may represent a suitable compromise to avoid the unnecessary proliferation of new names (Pickford, 1983). However, until such time as a detailed phylogenetic analysis has been undertaken, I continue to employ Hexaprotodon (sensu lato) , with the explicit understanding that it almost certainly represents a paraphyletic grouping of unrelated stem hippopo­tamines.

6.2. East Africa

6.2.1. Lothagam, Kanapoi, and the Baringo Basin

As discussed above, the majority of specimens from the Manonga Valley can be assigned to Hex. harvardi. The Manonga Valley material retains a series of primitive traits of the dentition that are characteristic of Hex. harvardi, but in combination are lost or modified in all other species of Hexaprotodon. These include the retention of six upper and lower incisors of subequal size, strongly bilaterally compressed lower canines, a high incidence of double-rooted upper and lower Pis, large premolars with well-developed accessory cusps and pustu­late crests, a premolar row that is sub equal in length to the molar row, brachyo­dont molars, and a less pronounced size differential in the molar series.

It is important to take into account that Hex. harvardi is derived from sites that span a considerable time range (ca. 7-4 Ma). Although the samples from different time periods and geographic areas are limited, the evidence suggests that Hex. harvardi represents part of an evolving lineage that underwent mor­phological change through time. The oldest specimens attributed to Hex. har­vardi, from the Mpesida Beds (6.4-7.0 Ma) in the Baringo basin, are morphologically very similar to the type material from Lothagam, but the teeth (and the postcranials) tend to be somewhat larger (the occlusal area of the cheek teeth is 16.6% larger on average, and more than one quarter of the teeth fall outside the 95% confidence limits of the sample from Lothagam). The younger material from Lukeino, which is apparently roughly contemporary with that from Lothagam (4.72-6.24 Ma), is morphologically and metrically indistinguish-

180 Chapter 6

able from the latter. Most significant in this regard, perhaps, is the collection from the site of Kana poi, which is younger than that fromLothagam and Lukeino, with an estimated age of 3.9-4.2 Ma (Leakey et al., 1995). The Kanapoi material displays a suite of more derived characteristics of the skull and dentition: the orbits are raised slightly above the dorsal surface of the neurocranium; the upper and lower central incisors are noticeably larger than the lateral pair of incisors; p1 is single-rooted; p4 is relatively small, with a poorly developed lingual cuspule; and the premolars are smaller in relation to the size of the molars. Detailed comparisons of the material from the Manonga Valley show that the dental samples from the Thole and Tinde Members are metrically and morpho­logically identical to those of Hex. harvardi from Lothagam and Lukeino, and that they are more conservative than the material from Kanapoi. This suggests that the Hex. harvardi samples from the lower beds of the Manonga Valley sequence are more comparable in age to Lothagam and Lukeino, at 5-6 Ma, than they are to the younger Kanapoi sample. There is some indication, however, that the isolated teeth and postcranials from the overlying Kiloleli Member are slightly larger and the molars more hypsodont that those from Tinde, and this may indicate the occurrence of a slightly more progressive form of Hex. harvardi in the Manonga Valley sequence. However, the differences are minor, and the samples too small to confirm this preliminary observation.

6.2.2. Turkana Basin

The Plio-Pleistocene hippopotamids from the Turkana basin have been re­viewed in some detail by Gaze (1985) and Harris (1991). The lower part of the sequence in the Dmo Valley (Shungura Members A and B) and Koobi Fora (lower Tulu Bor, Lokochot, Moiti, and Lonyumun Members of the Koobi Fora Forma­tion), which dates from at least 4.35 Ma to younger than 3 Ma (Harris et aI., 1988; Harris, 1991; Brown, 1994), has yielded abundant remains of a hexaprotodont hippopotamid, Hex. protamphibius (Arambourg, 1944a). Gaze (1985) referred this material to a separate subspecies, Hex. protamphibius turkanensis, in order to differentiate it from the younger tetraprotodont material belonging to Hex. protamphibius protamphibius (Arambourg, 1947). In addition, Gaze identified at least one other species of Hexaprotodon from the lower part of the Dmo sequence, Hex. shungurensis. However, I am inclined to agree with Harris (1991) that this latter species is possibly a junior synonym of Hex. protamphibius.

Hexaprotodon protamphibius turkanensis, which represents the earliest hip­popotamid to occur in the Turkana basin, is more derived than Hex. harvardi in the following respects: the sagittal crest and nuchal crest are raised above the level of the muzzle; the orbits are slightly more elevated, and are positioned further posteriorly (located along the length of the cranium about 60-70% from the tip of the snout); the central lower incisor is larger than the lateral pair, rather than sub equal; the premolars are smaller in relation to the size of the molars, and although variable, they have less of a tendency to develop accessory cuspules and pustulations; and the cheek teeth are smaller in overall size (they are on average less than 80% of the occlusal area of those in Hex. harvardi), and the lower molars are slightly more elongated (Harris et al., 1988; Harris, 1991). Since

Hippopotamidae 181

many of the differences seen in Hex. protamphibius represent a continuation of the evolutionary transformation seen in time-successive samples of Hex. har­vardi, it is reasonable to infer a direct ancestral-descendant relationship between them (Fig. 5).

6.2.3. Hadar

In addition to Trilobophorus afarensis discussed above, Gaze (1985) recog­nized a second species of hexaprotodont hippopotamid from the Hadar Forma­tion, Hex. coryndonae. Although poorly known, it appears to be more derived than Hex. harvardi, as well as other East African hippopotamids, in having 12 smaller than 13 (see Table VII).

6.2.4. Western Rift

A sizable collection of hippopotamids has been recovered from sites in Uganda and Zaire in the Western Rift, associated with the Lake Albert and Lake Edward basins (Hopwood, 1926; Cooke and Coryndon, 1970; Erdbrink and Krommenhoek, 1975; Pavlakis, 1987, 1990; Yasui et aI., 1992; Faure, 1994). At least two species are represented: a pygmy hexaprotodont, Hex. imagunculus (Hopwood, 1926), and a larger, more advanced hippopotamid, Hippopotamus kaisensis (Hopwood, 1926). Previous workers have recognized additional spe­cies. Cooke and Coryndon (1970), for instance, distinguished several isolated cranial and mandibular specimens, which they referred to Hippopotamus sp. Pavlakis (1987), following Gaze (1980), tentatively referred these specimens to ?Hex. shungurensis. However, given the lack of adequate material for compari­son, there seems little justification for distinguishing these specimens from Hex. imagunculus. Similarly, Erdbrink and Krommenhoek (1975) identified a new species, Hippopotamus (Choeropsisj arch ech oeropsis, supposedly somewhat smaller than Hex. imagunculus. However, as the task of correctly identifying the serial positions of cheek teeth is exceedingly difficult when using only isolated molars and jaw fragments of immature individuals, the metrical distinctiveness of these specimens may not be easy to verify. It seems equally plausible that these specimens can be attributed to Hex. imagunculus (see also Faure, 1994). In addition, Pickford (1990a) has recorded (without further explanation) additional species: Hippopotamus sp. from the late Miocene Oluka Formation, Hip. cf. aethiopicus from late Pliocene-early Pleistocene localities, and Hip. amphibius from the late Pleistocene Rwebishengo Beds. Also, Faure (1994) has recognized Hip. gorgops from the early Pleistocene Nyabusosi Formation (-1.8 Ma), and an undescribed species of Hippopotamus from the early Pliocene Nkondo Forma­tion (-4.5-5.0 Ma).

Much of the taxonomic confusion stems from the fact that until recently the hippopotamid material from Uganda and Zaire was very fragmentary, and that limited samples were available from sites of unknown age or of dubious strati­graphie association (Pickford, 1990a; Pickford et aI., 1988, 1991, 1992). The recent recovery of a sizable new collection of fossil hippotamids from the Western Rift (Faure, 1994), along with a substantially revised biostratigraphic scheme (Pickford et aI. 1992, 1993), may help alleviate some of these difficulties.

182 Chapter 6

Based on published material, there are two well-established species of hippo po­tamids from late Miocene and early Pliocene localities in the Western Rift, Hex. imagunculus and Hip. kaisensis. According to Faure (1994), both species occur in sediments ranging in age from about 5 Ma to 1.8 Ma. The possible occurrence of additional species, however, requires confirmation.

Hippopotamus kaisensis was originally proposed by Hopwood (1926) as a subspecies of Hip. amphibius, and although this designation has been followed by a number of workers (Hooijer, 1950; Erdbrink and Krommenhoek, 1975; Pavlakis, 1987, 1990), there are some distinctive morphological differences that merit the recognition of a separate species (Hopwood, 1939; Cooke and Coryn­don, 1970; Coryndon, 1978a; Geze, 1985; Faure, 1994). It is quite clear that this species is closely related to Hip. amphibius, and it probably represents the conservative sister taxon of both Hip. amphibius and Hip. gorgops (Fig. 5).

Hexaprotodon imagunculus can be distinguished from Hex. harvardi by the following specialized features: (1) it is considerably smaller in size (the com­bined area of the cheek teeth is only 56% that of Hex. harvardi); (2) the lower incisors are apparently hexaprotodont, but the two lateral incisors, especially 13 ,

are smaller than 110 not subequal; (3) the premolars are relatively small in comparison with the molars; (4) Pl is usually a single-rooted tooth; (5) the posterior premolars tend be less complex, with less prominent accessory cuspu­les; (6) the molars are slightly more hypsodont; and (7) the frontal does not make contact with the maxilla, due to the extension of the lacrimal medially (at least based on the Kazinga Channel partial cranium). Hex. imagunculus is more derived than Hex. harvardi, and it can readily be· distinguished on the basis of size and morphology from the material from the Manonga Valley. However, as Erdbrink and Krommenhoek (1975) have noted, there may be some justification for synonymizing Hex. imagunculus with Hex. hipponensis from North Africa.

6.3. North Africa

Several species of hippopotamid have been described from the late Miocene and early Pliocene of North Africa. Gaudry (1876a,b) described some isolated teeth of a small hippopotamid from Bone, Algeria, as a distinct species, Hippo­potamus (Hexaprotodon) hipponensis. There has been considerable discussion over whether or not the isolated incisors from the site indicate a hexaprotodont condition, but most workers have favored the original interpretation by Gaudry, that Hex. hipponensis had six lower incisors (Gaudry, 1876a,b; Pomel, 1890, 1896; Joleaud, 1920; Deperet, 1921; Arambourg, 1944b; Hooijer, 1950). Aram­bourg's (1944b) metrical data provide a ratio ofrelative incisor size of 100:79:76, in which 11 is somewhat larger than the lateral incisors, comparable to the condition in Hex. imagunculus. Similar material from Wadi Natrun in Egypt (which is probably similar in age to Bone, and correlated to MN 13), described by Andrews (1902) and Stromer (1914), has been referred to the same species, but the incisors, which are apparently associated, imply a tetraprotodont rather than hexaprotodont pattern. Based on this inferred difference, Arambourg (1947)

Hippopotamidae 183

argued that the material should be referred to Hex. protamphibius, although he distinguished it as a separate subspecies, andrewsi. Subsequent workers have generally considered the two samples to belong to a single species, Hex. hippo­nensis, although it may eventually prove necessary to recognize the species distinctiveness of Hex. andrewsi. Furthermore, as noted above, the cheek teeth from Bone and Wadi Natrun are morphologically and metrically very similar to Hex. imaguncu1us from the Western Rift. They are, however, distinct from Hex. harvardi in their smaller size and more specialized dentition.

In 1987, Gaziry described a new species of medium-size hexaprotodont hippopotamid, Hex. sahabiensis, from the early Pliocene locality of Sahabi in Libya (-5 Ma). Gaziry (1987) listed a number of features that purportedly distinguish Hex. sahabiensis from Hex. harvardi. However, his comparisons were based entirely on Coryndon's published descriptions and drawings of Hex. harvardi, and this led him to misrepresent the morphological differences be­tween the two species. For example, his claim that the development of accessory cusps, cingula, and enamel patterns on the molars and premolars is distinctive is of doubtful significance given the range of variability of these features in Hex. harvardi. Moreover, one of the major characteristics emphasized by Gaziry (1987) is the relatively much larger incisors and canines of Hex. sahabiensis. However, a closer inspection of Gaziry's data shows that he mistakenly compared Coryn­don's mesiodistal crown lengths of the incisors and canines of Hex. harvardi with the apicobasallengths of Hex. sahabiensis. It is hardly surprising, therefore, that he found the anterior teeth of his new species to be almost four times larger than the East African species! Nevertheless, Hex. sahabiensis is smaller (being more similar in size to Hex. aethiopicus and Hex. imaguncu1us) and more specialized than Hex. harvardi in having more reduced premolars and a greater size differential between the molars. These differences are probably adequate to distinguish Hex. sahabiensis from Hex. harvardi at the species level. However, the taxonomic distinctiveness of Hex. sahabiensis in relation to the contempo­rary North African species, Hex. hipponensis and Hex. andrewsi, which are closely comparable in size and morphology, as well as Hex. crusa/onti and Hex. primaevus from Spain, needs further investigation. Also pertinent here may be an undescribed Hexaprotodon from the Baynunah Formation of Abu Dhabi on the Arabian peninsula, which is probably of similar age (Hill et a1., 1990; Whybrow et a1., 1990).

6.4. Europe and Asia

Toward the end of the Miocene, during the late Messinian, at sites correlated with MN 13, Hexaprotodon appears for the first time in the fossil record of Eurasia. It is part of a limited exchange of large mammals between Africa and Eurasia at this time, which also included the arrival from Africa of macaques and reduncine bovids. Hexaprotodon appears to have established a wide geo­graphic range-extending from Spain in western Europe to Myanmar in Asia­within a very short period oftime after its initial arrival. This may simply reflect

184 Chapter 6

the availability in tropical and subtropical Eurasia of a previously unoccupied niche for large semiaquatic herbivores, although the timing and pattern of dispersion into Eurasia is consistent with a model in which the rapid geographic expansion and diversification of Hexaprotodon in Africa just prior to its arrival in Eurasia was merely extended beyond the geographic limits of Africa.

The best known species from the late Miocene or early Pliocene of Eurasia is Hexaprotodon sivalensis from the Siwalik Hills of India and Pakistan (Falconer and Cautley, 1836; Lydekker, 1884; Colbert, 1935; Hooijer, 1950). Barry and Flynn (1989) record the earliest occurrence of this species in the Siwaliks at the Mio-Pliocene boundary, at 5.3 Ma. It is comparable in size and morphology to Hex. harvardi, and, in fact, in many respects, it appears to be much more similar to Hex. harvardi than are contemporary species in North Africa. However, it is more derived in a number of features: the orbits are more elevated; the lateral incisors tend to be slightly reduced relative to the size of the central incisor; the premolars are relatively small in relation to the size of the molars, and not as pustulate; and the lower molars exhibit a greater size differential as they progress posteriorly. In addition, it is worth noting that the cheek teeth of Hex. sivalensis show a remarkable degree of wear, even in quite young individuals, when compared with Hex. harvardi. and this may further imply an important ecobe­havioral distinction between the two species. A second Asian species, Hex. iravaticus (Falconer and Cautley. 1847), slightly smaller, and possibly less derived than Hex. sivalensis, has been recovered from the early Pliocene Upper Irrawaddy Beds in Myanmar (Falconer and Cautley, 1836; Lyddeker, 1884; Colbert, 1938, 1943; Hooijer, 1950). However, it is poorly known, making it difficult to assess its taxonomic and phylogenetic relationships.

In Europe, species of Hexaprotodon have been recovered from several late Miocene (MN 13) sites in Italy, Spain. and France. Pantanelli (1879) described some fragmentary remains of a small hexaprotodont hippopotamid from Casino in Italy, which he referred to Hip. hipponensis. Later, Joleaud (1920) erected a new species, Hip. pantanellii. The material is similar in size and is consistent in morphology with the contemporary Hex. iravaticus from Asia and Hex. hippo­nensis from North Africa, although most authors have tended to maintain its distinctiveness (Joleaud, 1920; Deperet, 1921; Hooijer, 1946, 1950; Erdbrink and Krommenhoek, 1975; Coryndon, 1978a). A hippopotamid similar in size to Hex. pantanellii has also been described from Gravitelli in Sicily (Seguenza, 1902, 1907). Although the original specimens were destroyed in the Messina earth­quake of 1908, Hooijer (1946) used Seguenza's published descriptions and figures to assign the material to a separate species, Hip. siculus.

In Spain, fragmentary remains of a medium-size species of hippopotamid have been recovered from a number of localities correlated with MN 13 (Morales, 1981; Mein, 1989; Moya-Sola and Agustf, 1989). Aguirre (1963) described a new species, Hip. crusafonti, on the basis of material from Arenas del Rey, and Morales (1981) has assigned additional material from La Portera. In addition, comparable material of similar age, and probably referrable to the same species, has also been recovered from Arquillo and Venta del Moro in Spain (Crusafont et aI., 1963, 1964; Morales, 1981) and Mosson in France (Faure and Meon, 1984;

Hippopotamidae 185

Faure, 1985). Crusafont et al. (1963), however, have given the name Hip. (Hex.) primaevus to the Arquillo specimens, but since no holotype was identified, it is considered a nomen nudum. Although poorly· known, Hex. crusafonti is a relatively specialized hippopotamid, with only four incisors in the mandible (it is, in fact, the earliest known tetraprotodont), relatively small, simple premolars, and narrow, hypsodont molars. It is probably derived from a form such as Hex. hipponensis, and appears to have close affinities with Hex. siculus from Sicily, with which it may eventually prove to be synonymous (Fig. 5).

7. Summary and Conclusions

Based on detailed comparisons of the Manonga Valley material with fossil hippopotamids from other Miocene and Plio-Pleistocene sites in Africa and Eurasia, the following conclusions can be reached: (1) most of the specimens from the Manonga Valley can be referred to Hex. harvardi; the material retains a series of primitive traits of the dentition and cranium that are characteristic of Hex. harvardi, but in combination are lost or modified in all other species of hippopotamines; (2) the specimens from the Ibole and Tinde Members are morphologically and metrically indistinguishable from Hex. harvardi from Lothagam (the type site) and the Lukeino Formation, thereby implying a biostra­tigraphic age of 5-6 Ma; (3) the specimens from the Kilolei Member are referred to Hex. harvardi, but comparisons suggest the possibility that they may represent a slightly more progressive form than that from the Tinde and Ibole Members; (4) a single isolated phalanx appears to indicate the occurrence of a second, smaller species of hexaprotodont hi ppopotamid in the Manonga Valley; (5) Hex. harvardi is apparently restricted to late Miocene and early Pliocene sites in East Africa, dated to between 7.0 and 4.0 Ma; (6) the Manonga Valley material serves to extend the geographic range of the species from southern Ethiopia and northern Kenya (principally the Thrkana and Baringo basins) southward as far as northern Tanzania; (7) the postcranial skeleton of Hex. harvardi is lightly built, with relatively long and slender limb bones and cheirida, suggesting a more digitigrade, fast-running, cursorial habit, in contrast to that of the more am­phibiously adapted Hippopotamus; (8) Hex. harvardi is the earliest known member of the Hippopotaminae, and, based on its conservative craniodental morphology, the species appears to represent the primitive sister taxon of all other hippopotamines (with the possible exception of Choeropsis liberiensis); (9) the genus Hexaprotodon, as currently perceived, is a paraphyletic clustering of stem hippopotamines that lacks derived features of Hippopotamus, a taxo­nomic issue that can only be resolved with more detailed comparative analyses; and (10) the living pygmy hippopotamus, Choeropsis liberiensis, appears to be more primitive still than Hex. harvardi and should be retained in a separate genus (contra Coryndon 1977, 1978a).

NOTE ADDED IN PROOF. In 1996 WMPE recovered additional hippopotamids from Tinde, including several specimens that contribute pertinent new information

186 Chapter 6

on the anatomy and taxonomic placement of Hexaprotodon harvardi from the Manonga Valley. The most important specimens are as follows:

WM 696/96, from Tinde East, consists of a right premaxilla fragment with 11_13• The specimen confirms the general morphology of the anterior lower face and the hexaprotodont nature of the incisors, as seen in 056/90. The three incisors are subequal in size, with relative proportions of 98 : 100 : 91 (see Table IV), and they are arranged in a gently curving arc. The estimated breadth of the palate across the premaxilla is 120 mm. This specimen provides further confir­mation of the close similarity of the Tinde hippopotamid to Hexaprotodon harvardi from Lothagam.

WM 1100/96 consists of a partial cranium of a juvenile, which represents the most complete specimen of a hippopotamid so far recovered from the Manonga Valley. It was discovered at Tinde West by Dr. Avelin Malyango. The specimen preserves much of the maxilla, portions of the frontal, nasal, lacrimal and temporal bones on the right side, a fragment of the occiput, as well as left C, p1, dp3 , Ml - 3 and right C, dp2- 3 , Ml - 2 • The following combination of primitive features confirms its assignment to Hex. harvardi: (1) the muzzle is short in relation to the length of the neurocranium; (2) the orbits are situated low on the face, and they do not extend superiorly above the roof of the muzzle; (3) the anterior root of the zygomatic arch is stout, widely flaring in relation to the canine flanges, and situated very low on the face; (4) the nasal and lacrimal bones are separated by a well-developed antorbital process of the frontal; and (5) the p1 is a large, single-cusped tooth with a bilobate or bifid root.

ACKNOWLEDGMENTS. I am grateful to the following individuals and institutions for their help and access to specimens: Wolfgang Fuchs, Ross MacPhee, and Guy Musser of the American Museum of Natural History, New York; Peter Andrews and Jerry Hooker of the Natural History Museum, London; Frederick Kyalo Manthi, Nina Mudida, and Mary Muungu of the National Museums of Kenya, Nairobi; William Bongo, Amandus Kweka, and Michael Mbago of the National Museums of Tanzania, Dar es Salaam. Special thanks go to Alan Gentry, John Harris, and Meave Leakey for discussions on the anatomy, systematics, and evolutionary history of hippo po tam ids. Ray Bernor, Laura Bishop, Andrew Hill, Tania Mirza, Bill Sanders, and Jacques Verniers contributed ideas and made helpful suggestions to improve the manuscript. I thank the Office of the President and the Tanzania Commission for Science and Technology for permission to conduct research in Kenya and Tanzania, respectively. This research was sup­ported by grants from the Boise Fund, the National Geographic Society, the L.S.B. Leakey Foundation, and the National Science Foundation (DBS-9120882).

References

Aguirre, E. de, 1963, Hippopotamus crusafonti n. sp. del Plioceno inferior de Arenas del Rey (Granada), Notas Comun., Madrid, 69:215-230.

Andrews, C. W., 1902, Note on a Pliocene vertebrate fauna from the Wadi-Natrun, Egypt, Geol. Mag., London, n.s., Decade 4, 9:433-439.

Hippopotamidae 187

Arambourg, C., 1944a, Les Hippopotames fossiles d'Afrique, C. R. Acad. Sci. Paris 218:602-604. Arambourg, C., 1944b, Au sujet de I'Hippopotamus hipponensis Gaudry, Bull. Soc. Geol. France,

Ser. 5, 14:147-152. Arambourg, C., 1947, Contribution a l'etude geologique et paleontologique du bassin du lac Rudolphe

et de la basse vallee de l'Omo, Mission scientifique de l'Omo 1932-1933, Volume I: Geologie et Anthropologie, pp. 231-562.

Barry, J., and Flynn, 1., 1989, Key biostratigraphic events in the Siwalik sequence, in: European Neogene Mammal Chronology (E. H. Lindsey, V. Fahlbusch, and P. Mein, eds.l, Plenum Press, New York, pp. 557-571.

Behrensmeyer, A. K., 1976, Lothagam Hill, Kanapoi and Ekora: A general summary of stratigraphy and faunas, in: Earliest Man and Environments in the Lake Rudolf Basin: Stratigraphy. Paleoe­cology. and Evolution (Y. Coppens, F. C. Howell, G. 1. Isaac, and R. E. F. Leakey, eds.l, University of Chicago Press, Chicago, pp. 163-170.

Bishop, W. w., and Chapman, G. R., 1970, Early Pliocene sediments and fossils from the northern Kenya Rift Valley, Nature 226:914-918.

Bishop, W. W., Hill, A., and Miller, J. A., 1971, Succession of Cainozoic vertebrate assemblages from the northern Kenya Rift Valley, Nature 233:389-394.

Bishop, W. W., and Pickford, M. H. 1., 1975, Geology, fauna and palaeoenvironments of the Ngorora Formation, Kenya Rift Valley, Nature 254:185-192.

Boaz, N. T., 1994, Significance ofthe Western Rift for hominid evolution, in: Integrative Paths to the Past: Paleoanthropological Advances in Honor of F. Clark Howell (R. S. Corruccini and R. 1. Ciochon, eds.l, Prentice Hall, Englewood Cliffs, NJ, pp. 321-343.

Brown, F. H., 1994, Development of Pliocene and Pleistocene chronology of the Turkana Basin, East Africa, and its relation to other sites, in: Integrative Paths to the Past: Paleoanthropological Advances in Honor of F. Clark Howell (R. S. Corruccini and R. 1. Ciochon, eds.l, Prentice Hall, Englewood Cliffs, NJ, pp. 285-312.

Colbert, E. H., 1935, Siwalik mammals in the American Museum of Natural History, Trans. Amer. Phil. Soc. 26:1-401.

Colbert, E. H., 1938, Fossil mammals from Burma in the American Museum of Natural History, Bull. Amer. Mus. Nat. Hist. 74:255-436.

Colbert, E. H., 1943, Pleistocene vertebrates collected in Burma by the American Southeast Asiatic Expedition, Trans. Amer. Philos. Soc. 32:395-429.

Cooke, H. B. S., and Coryndon, S. C., 1970, Pleistocene mammals from the Kaiso Formation and other related deposits in Uganda, in: Fossil Vertebrates of Africa, Vol. 2 (1. S. B. Leakey and R. J. G. Savage, eds.l, Academic Press, London, pp. 107-224.

Coryndon, S. C., 1970, The extent of variation in fossil Hippopotamus from Africa, Symp. Zool. Soc. Lond.26:135-147.

Coryndon, S. C., 1976, Fossil Hippopotamidae from the Plio-Pleistocene successions of the Rudolf Basin, in: Earliest Man and Environments in the Lake Rudolf Basin (Y. Coppens, F. C. Howell, G. Ll. Isaac, and R. E. F. Leakey, eds.l, University of Chicago Press, Chicago, pp. 238-250.

Coryndon, S. c., 1977, The taxonomy and nomenclature of the Hippopotamidae (Mammalia, Artiodactylal and a description of two new fossil species, Proc. Kon. Ned. Akad. Wetensch. B80:61-88.

Coryndon, S. C., 1978a, Hippopotamidae, in: Evolution of African Mammals (V. J. Maglio and H. B. S. Cooke, eds.). Harvard University Press, Cambridge, pp. 483-495.

Coryndon, S. C., 1978b, Fossil Hippopotamidae from the Baringo Basin and relationships within the Gregory Rift, Kenya, in: Geological Background to Fossil Man (w. W. Bishop, ed.l, Scottish Academic Press, Edinburgh, pp. 279-292.

Coryndon, S. C., and Coppens, Y., 1973, Preliminary report on Hippopotamidae (Mammalia, Artiodactylal from the Plio/Pleistocene of the lower Omo Basin, Ethiopia, in: Fossil Vertebrates of Africa, Vol. 3 (1. S. B. Leakey, R. J. G. Savage, and S. C. Coryndon, eds.l, Academic Press, London, pp. 139-157.

Coryndon, S. C., and Coppens, Y., 1975, Une espece nouvelle d'Hippopotame nain Plio-Pleistocene du bassin du lac Rodolphe (Ethiopie, Kenya). C. R. Acad. Sci. Paris 280:1777-1780.

188 Chapter 6

Crusafont, M., Adrover, R., and Golpe, J. M., 1963, Hallazgo en Espana del mas primitivo Hipopotamo fosil conocido hasta la actualidad, Notas Comun. 69:71-79.

Crusafont, M., Adrover, R., and Golpe, J. M., 1964, D~coverte dans Ie Pikermien d'Espagne du plus primitif des Hippopotames: Hippopotamus (Hexaprotodon) primaevus n. sp. C. R. Acad. Sci. Paris 258:1572-1575.

Dep~ret, C., 1921, La phylog~nie des hippopotames, Bull. Soc. Geol. France S~r. 4 21:163-165. Erdbrink, D., and Krommenhoek, W., 1975, Contribution to the knowledge of the fossil Hippopo­

tamidae from the Kazinga Channel area (Uganda), Sauget. Mitt. Miinchen 4:258-294. Falconer, H., and Cautley, P. T., 1836, Note on the fossil Hippopotamus of the Sivalik Hills, Asiatic

Res. 19:39-53. Falconer, H., and Cautley, P. T., 1847, Fauna Antiqua Sivalensis, being the Fossil Zoology of the

Sewalik Hills, in the North of India, R. Hardwicke, London. Faure, M., 1985, Les Hippopotames Quaternaires non-insulaires d'Europe Occidentale, Nouv. Arch.

Mus. Hist. Nat. Lyon 23:13-79. Faure, M., 1994, Les Hippopotamidae (Mammalia, Artiodactyla) du rift occidental (Bassin du lac

Albert, Ouganda). Etude pr~liminaire, in: Geology and Palaeobiology of the Albertine Rift Valley Uganda-Zaire, Vol. II: Palaeobiology (B. Senut and M. Pickford, eds.), ClFEG Occasional Publi­cation 29, CIFEG, Orl~ans, pp. 321-337.

Faure, M., and M~on, H., 1984, I.:Hippopotamus crusafonti de La Masson (pres Montpellier) premier reconnaissance d'un hippopotame n~og~ne en France, C. R. Acad. Sci. Paris, S~r. II, 298:93-98.

Gaudry, A., 1876a, Sur un Hippopotame fossile d~couvert a Bone (Alg~rie), Bull. Soc. GeoJ. France 3, S~r. 4:501-504.

Gaudry, A., 1876b, Sur un Hippopotame a six incisives inferieures trouv~ fossile en Algerie, C. R. Acad. Sci. Paris 83:90-92.

Gaziry, A. W., 1987, Hexaprotodon sahabiensis (Artiodactyla, Mammalia): A new hippopotamus from Libya, in: Neogene Paleontology and Geology of Sahabi (N. T. Boaz, A. EI-Arnauti, A. W. Gaziry, J. de Heinzelin and D. D. Boaz, eds.), Alan R. Liss, New York, pp. 303-315.

Geze, R., 1980, Les Hippopotamidae (Mammalia, Artiodactyla) du Plio-Pleistocime de l'Ethiopia (Afrique Orientale), These 3e cycle, Universite de Paris, pp. 1-116.

Geze, R., 1985, R~partition paleo~cologique et relations phylog~n~tiques des Hippopotamidae (Mammalia, Artiodactyla) du N~og~ne d' Afrique Orientale, in: I:environnement des Hominides au Plio-PIeistoci!me (Y. Coppens, ed.), Masson, Paris, pp. 81-100.

Harris, J. M., 1991, Family Hippopotamidae, in: Koobi Fora Research Project, Volume 3. The Fossil Ungulates: Geology. Fossil Artiodactyls, and Palaeoenvironments (J. M. Harris, ed.), Clarendon Press, Oxford, pp. 31-85.

Harris, J. M., Brown, F. H., and Leakey, M. G., 1988, Stratigraphy and paleontology of Pliocene and Pleistocene localities west of Lake Turkana, Kenya, Contributions in Science, Natural History Museum of Los Angeles County, No. 399, pp. 1-128.

Harrison, T., 1993, The anatomy, paleobiology and phylogenetic relationships of Hexaprotodon harvardi (Mammalia, Hippopotamidae) from the Neogene of East Africa, I. Vert. Paleontol. 13 (Abstracts}:40A.

Harrison, T., and Verniers, J., 1993, Preliminary study of the stratigraphy and mammalian palaeon­tology of Neogene sites in the Manonga Valley, northern Tanzania, N. lb. Geol. Palaont. Abh. 190:57-74.

Hill, A., 1994, Late Miocene and early Pliocene hominoids from Africa, in: Integrative Paths to the Past: Paleoanthropological Advances in Honor of F. Clark Howel1 (R. S. Corruccini and R. L. Ciochon, eds.), Prentice Hall, Englewood Cliffs, NJ, pp. 123-145.

Hill, A., and Ward, S., 1988, Origin of the Hominidae: The record of African large hominoid evolution between 14 My and 4 My, Yrbk Phys. AnthropoJ. 31:49-83.

Hill, A., Drake, R., Tauxe, L., Monaghan, M., Barry, J. C., Behrensmeyer, A. K., Curtis, G., Fine Jacobs, B., Jacobs, L., Johnson, N., and Pilbeam, D., 1985, Neogene palaeontology and geochronology of the Baringo Basin, Kenya, I. Hum. Evo!. 14:759-773.

Hill, A., Curtis, G., and Drake, R., 1986, Sedimentary stratigraphy ofthe Tugen Hills, Baringo, Kenya, in: Sedimentation in the African Rift System (L. E. Frostick, R. Renaut, I. Reid, and J-J. TIercelin, eds.), Blackwell, Oxford, pp. 285-295.

Hippopotamidae 189

Hill, A., Whybrow, P., and Yasin al-Tikriti, W., 1990, Late Miocene primate fauna from the Arabian Peninsula: Abu Dhabi, United Arab Emirates, Am. J. Phys. Anthrop01. 81:240-241.

Hill, A., Ward, S., and Brown, B., 1992, Anatomy and age of the Lothagam mandible, J. Hum. Ev01. 22:439-451.

Hooijer, D. A., 1946, Notes on some Pontian mammals from Sicily figured by Seguenza, Arch. Neerl. Zool. 7:301-333.

Hooijer, D. A., 1950, The fossil Hippopotamidae of Asia, with notes on the recent species, Zool. Verhandel. 8:1-124.

Hopwood, A. T., 1926, Fossil Mammalia, in: The Geology and Palaeontology of the Kaiso Bone Beds (E. J. Wayland, ed.), Dcc. Pap. Geol. Surv. Uganda 2:13-36.

Hopwood, A. T., 1939, The mammalian fossils, in: The Prehistory of Uganda Protectorate (T. P. O'Brien, ed.), University of Cambridge Press, Cambridge, pp. 308-316.

Joleaud, L., 1920, Contribution a l'etude des Hippopotames fossiles, Bull. Soc. Geol. France, Ser. 4 20:13-26.

Kalb, J. E., 1993, Refined stratigraphy ofthe hominid-bearing Awash Group, Middle Awash Valley, Afar Depression, Ethiopia, Newsl. Stratigr. 29:21-62.

Kalb, J. E., and Jolly, C. J., 1982, Late Miocene and early Pliocene formations in the Middle Awash Valley, Ethiopia, and their bearing on the zoogeography of Sahabi, Garyounis Sci. BulI. (Libya), Special Issue 4:123-132.

Kalb, J. E., and Mebrate, A., 1993, Fossil elephantoids from the hominoid-bearing Awash Group, Middle Awash Valley, Afar Depression, Ethiopia, 'frans. Am. Phil. Soc. 83:1-114.

Kalb, J. E., Jolly, C. J., Mebrate, A., Tebedge, S., Smart, C., Oswald, E. B., Cramer, D., Whitehead, P., Wood, C. B., Conroy, G. C., Adefris, T., Sperling, L., and Kana, B., 1982a, Fossil mammals and artefacts from the Middle Awash Valley, Ethiopia, Nature 298:25-29.

Kalb, J. E., Jolly, C. J., Tebedge, S., Mebrate, A., Smart, C., Oswald, E. B., Whitehead, P. F., Wood, C. B., Adefris, T., and Rawn-Schatzinger, V., 1982b, Vertebrate faunas from the Awash Group, Middle Awash Valley, Afar, Ethiopia, J. Vert. Paleontol. 2:237-258.

Kalb, J. E., Oswald, E. B., Tebedge, S., Mebrate, A., Tola, E., and Peak, D., 1982c, Geology and stratigraphy of Neogene deposits, Middle Awash Valley, Afar, Ethiopia, Nature 298:17-25.

Leakey, M. G., Feibel, C. S., Bernor, R. 1., Harris, J. M., Cerling, T. E., Stewart, K. M., Storrs, G. w., Walker, A., Werdelin, 1., and Winkler, A. J., 1996, Lothagam: A record of faunal change in the late Miocene of East Africa, J. Vert. Paleontol. 16:556-570.

Leakey, M. G., Feibel. C. S., MacDougal. I., and Walker, A., 1995, New four-million-year-old hominid species from Kanapoi and Allia Bay, Kenya, Nature 376:565-571.

Lydekker, R., 1884, Siwalik and Narbada bunodont suina, Mem. Geol. Surv. India, Ser. 10,3:35-104. Maglio, V. J., 1974, Late Tertiary fossil vertebrate successions in the northern Gregory Rift, East Africa,

Ann. Geol. Surv. Egypt 4:269-286. Mein, P., 1989, Updating of MN Zones, in: European Neogene Mammal Chronology (E. H. Lindsey,

V. Fahlbusch, and P. Mein, eds.), Plenum Press, New York, pp. 73-90. Morales, J., 1981, Venta del Moro: su macrofauna de mamiferos. y biostratigrafia continental del

Mioceno terminal Mediterraneo, Tesis Doctoral, Facultad de Geologia, Departmento de Paleon­tologia, Universidad Complutense de Madrid.

Moya-Sola, S., and Agusti, J., 1989, Bioevents and mammal successions in the Spanish Miocene, in: European Neogene Mammal Chronology (E. H. Lindsey, V. Fahlbusch, and P. Mein, eds.), Plenum Press, New York, pp. 357-373.

Pantanelli, D., 1879, Sugli strati miocenici del Casino (siena) e considerazioni suI miocene superiore. Att. R. Accad. Lincei, Ann. 276, Ser. III, 3:309-327.

Patterson, B., 1966, A new locality for early Pleistocene fossils in north-western Kenya, Nature 212:577-581.

Patterson, B., Behrensmeyer, A. K., and Sill, W. D., 1970, Geology and fauna of a new Pliocene locality in north-western Kenya, Nature 226:918-921.

Pavlakis, P. P., 1987, Biochronology. Paleoecology. and Biogeography of the Plio-Pleistocene Western Rift Fossil Mammal Faunas and Implications for Hominid Evolution, Ph. D. thesis, Department of Anthropology, New York University.

190 Chapter 6

Pavlakis, P. P., 1990, Plio-Pleistocene Hippopotamidae from the Upper Semliki, in: Results from the Semliki Research Expedition (N. T. Boaz, ed.), Virginia Museum of Natural History Memoir, No. 1, pp. 203-223.

Pickford, M., 1975, Late Miocene sediments and fossils from the northern Kenya Rift Valley, Nature 256:279-284.

Pickford, M., 1978a, Geology, palaeoenvironments and vertebrate faunas of the mid-Miocene Ngorora Formation, Kenya, in: Geological Background to Fossil Man (w. W. Bishop, ed.), Edinburgh, Scottish Academic Press, pp. 237-262.

Pickford, M., 1978b, Stratigraphy and mammalian palaeontology of the late Miocene Lukeino Formation, Kenya, in: Geological Background to Fossil Man (W. W. Bishop, ed.), Edinburgh, Scottish Academic Press, pp. 263-278.

Pickford, M., 1983, The origins of Hippopotamidae together with descriptions of two new species, a new genus and a new subfamily from the Miocene of Kenya, Geobios 16:193-217.

Pickford, M., 1989, Update on hippo origins, C. R. Acad. Sci. Paris 309, Ser. II:163-168. Pickford, M., 1990a, Uplift of the roof of Africa and its bearing on the evolution of mankind, Hum.

Evol. 5:1-20. Pickford, M., 1990b, Decouverte de Kenyapotamus en Tunisie, Ann. PaJeont. 76:277-283. Pickford, M., Senut, B., Ssemmanda, I., Elepu, D., and Obwona, P., 1988, Premiers resultats de la

mission de l'Uganda Palaeontology Expedition a Nkondo (Pliocene du Bassin Lac Albert, Ouganda), C. R. Acad. Sci. Paris 306, Ser. II:315-320.

Pickford, M., Senut, B., Roche, H., Mein, P., Ndaati, G., Obwona, P., and Tuhumwire, J., 1989, Uganda Palaeontology Expedition: resultats de la deuxieme mission (1987) dans la region de Kisegi­Nyabusosi (Bassin du Lac Albert, Ouganda), C. R. Acad. Sci. Paris Ser. II, 308:1751-1758.

Pickford, M., Senut, B., Vincens, A., Van Neer, W., Ssemmanda, I., Baguma, Z., and Musiime, E., 1991, Nouvelle biostratigraphie du Neogene et du Quaternaire de la region de Nkondo (Bassin du lac Albert, Rift occidental ougandais). Apport a l'evolution des paleomilieux, C. R. Acad. Sci Paris ser. II, 312:1667-1672.

Pickford, M., Senut, B., Ambrosi. J-P., Dechamps, R., Faure, M., Van Damme, D., Texier, p-J., Baguma, Z., and Musiime, E., 1992, Revision de la biostratigraphie du Neogeme du Rift Occidental (Ouganda-ZaYre), C. R. Acad. Sci. Paris Ser. II, 315:1289-1292.

Pickford, M., Senut, B., and Hadoto, D., 1993, Geology and Palaeobiology of the Albertine Rift Valley Uganda-Zaire, Vol. I: Geology, ClFEG Occasional Publication 24, ClFEG, Orleans.

Pomel, A., 1890, Sur les Hippopotames fossiles de l'Algerie, C. R. Acad. Sci. Paris 110:1112-1116. Pomel, A., 1896, Sur les Hippopotames fossiles de l'Algerie, C. R. Acad. Sci. Paris 123:1241-1242. Seguenza, L., 1902, I vertebrati fossili della Provincia di Messina. Parte Seconda. Mammiferi e

geologia del Piano Pontico, Boll. Soc. Geol. Ital. 21:115-175. Seguenza, L., 1907, Nuovi resti di mammiferi Pontici di Gravitelli presso Messina, Boll. Soc. Geol.

Ital. 26:89-122. Smart, C., 1976, The Lothagam 1 fauna: Its phylogenetic, ecological, and biogeographic significance,

in: Earliest Man and Environments in the Lake Rudolf Basin: Stratigraphy. Paleoecology. and Evolution, (Y. Coppens, F. C. Howell, G. L. Isaac, and R. E. F. Leakey, eds.), University of Chicago Press, Chicago, pp. 361-369.

Steunes, S., 1981, On the bi-rooted upper first premolar in some specimens of the Madagascar Holocene Hippopotamidae, Bull. Geol. Inst. Univ. Uppsala 9:30-32.

Steunes, S., 1989, Taxonomy, habits, and relationships of the subfossil Madagascan hippopotami Hippopotamus lemerlei and H. madagascariensis, J. Vert. Paleontol. 9:241-268.

Stromer, E., 1914, Mitteilungen iiber Wirbeltierreste aus dem Mittelpliocan des Natrontales (Agyp­ten). 3. Artiodactyla: A. Bunodontia: Flusspferd, Z. Deutschen Geol. Ges. 66:1-33.

Whybrow, P. J., Hill, A., Yasin al-TIkriti, w., and Hailwood, E. A., 1990, Late Miocene primate fauna, flora and initial palaeomagnetic data from the Emirate of Abu Dhabi, United Arab Emirates, J. Hum. Evol. 19:583-588.

Yalden, D. W., 1971, The functional morphology of the carpus in ungulate mammals, Acta Anat. 78:461-487.

Yasui, K., Kunimatsu, Y., Kuga, N., Bajope, B., and Ishida, H., 1992, Fossil mammals from the Neogene strata in the Sinda basin, eastern Zaire, African Study Monogr., Kyoto Univ., Suppl. 17:87-107.